>> Announcer: From Denver, Colorado, it's theCUBE, covering Super Computing 17. Brought to you by Intel. (soft electronic music) Hey, welcome back, everybody. Jeff Frick here with theCUBE. We're gettin' towards the end of the day here at Super Computing 2017 in Denver, Colorado. 12,000 people talkin' really about the outer limits of what you can do with compute power and lookin' out into the universe and black holes and all kinds of exciting stuff. We're kind of bringin' it back, right? We're all about democratization of technology for people to solve real problems. We're really excited to have our last guest of the day, bringin' the energy, Armughan Ahmad. He's SVP and GM, Hybrid Cloud and Ready Solutions for Dell EMC, and a many-time CUBE alumni. Armughan, great to see you. >> Yeah, good to see you, Jeff. So, first off, just impressions of the show. 12,000 people, we had no idea. We've never been to this show before. This is great. >> This is a show that has been around. If you know the history of the show, this was an IEEE engineering show, that actually turned into high-performance computing around research-based analytics and other things that came out of it. But, it's just grown. We're seeing now, yesterday the super computing top petaflops were released here. So, it's fascinating. You have some of the brightest minds in the world that actually come to this event. 12,000 of them. >> Yeah, and Dell EMC is here in force, so a lot of announcements, a lot of excitement. What are you guys excited about participating in this type of show? >> Yeah, Jeff, so when we come to an event like this, HBC-- We know that HBC is also evolved from your traditional HBC, which was around modeling and simulation, and how it started from engineering to then clusters. It's now evolving more towards machine learning, deep learning, and artificial intelligence. So, what we announced here-- Yesterday, our press release went out. It was really related to how our strategy of advancing HBC, but also democratizing HBC's working. So, on the advancing, on the HBC side, the top 500 super computing list came out. We're powering some of the top 500 of those. One big one is TAC, which is Texas Institute out of UT, University of Texas. They now have, I believe, the number 12 spot in the top 500 super computers in the world, running an 8.2 petaflops off computing. >> So, a lot of zeros. I have no idea what a petaflop is. >> It's very, very big. It's very big. It's available for machine learning, but also eventually going to be available for deep learning. But, more importantly, we're also moving towards democratizing HBC because we feel that democratizing is also very important, where HBC should not only be for the research and the academia, but it should also be focused towards the manufacturing customers, the financial customers, our commercial customers, so that they can actually take the complexity of HBC out, and that's where our-- We call it our HBC 2.0 strategy, off learning from the advancements that we continue to drive, to then also democratizing it for our customers. >> It's interesting, I think, back to the old days of Intel microprocessors getting better and better and better, and you had Spark and you had Silicon Graphics, and these things that were way better. This huge differentiation. But, the Intel I32 just kept pluggin' along and it really begs the question, where is the distinction now? You have huge clusters of computers you can put together with virtualization. Where is the difference between just a really big cluster and HBC and super computing? >> So, I think, if you look at HBC, HBC is also evolving, so let's look at the customer view, right? So, the other part of our announcement here was artificial intelligence, which is really, what is artificial intelligence? It's, if you look at a customer retailer, a retailer has-- They start with data, for example. You buy beer and chips at J's Retailer, for example. You come in and do that, you usually used to run a SEQUEL database or you used to run a RDBMS database, and then that would basically tell you, these are the people who can purchase from me. You know their purchase history. But, then you evolved into BI, and then if that data got really, very large, you then had an HBC cluster, would which basically analyze a lot of that data for you, and show you trends and things. That would then tell you, you know what, these are my customers, this is how many times they are frequent. But, now it's moving more towards machine learning and deep learning as well. So, as the data gets larger and larger, we're seeing datas becoming larger, not just by social media, but your traditional computational frameworks, your traditional applications and others. We're finding that data is also growing at the edge, so by 2020, about 20 billion devices are going to wake up at the edge and start generating data. So, now, Internet data is going to look very small over the next three, four years, as the edge data comes up. So, you actually need to now start thinking of machine learning and deep learning a lot more. So, you asked the question, how do you see that evolving? So, you see an RDBMS traditional SQL evolving to BI. BI then evolves into either an HBC or hadoop. Then, from HBC and hadoop, what do you do next? What you do next is you start to now feed predictive analytics into machine learning kind of solutions, and then once those predictive analytics are there, then you really, truly start thinking about the full deep learning frameworks. >> Right, well and clearly like the data in motion. I think it's funny, we used to make decisions on a sample of data in the past. Now, we have the opportunity to take all the data in real time and make those decisions with Kafka and Spark and Flink and all these crazy systems that are comin' to play. Makes Hadoop look ancient, tired, and yesterday, right? But, it's still valid, right? >> A lot of customers are still paying. Customers are using it, and that's where we feel we need to simplify the complex for our customers. That's why we announced our Machine Learning Ready Bundle and our Deep Learning Ready Bundle. We announced it with Intel and Nvidia together, because we feel like our customers either go to the GPU route, which is your accelerator's route. We announced-- You were talking to Ravi, from our server team, earlier, where he talked about the C4140, which has the quad GPU power, and it's perfect for deep learning. But, with Intel, we've also worked on the same, where we worked on the AI software with Intel. Why are we doing all of this? We're saying that if you thought that RDBMS was difficult, and if you thought that building a hadoop cluster or HBC was a little challenging and time consuming, as the customers move to machine learning and deep learning, you now have to think about the whole stack. So, let me explain the stack to you. You think of a compute storage and network stack, then you think of-- The whole eternity. Yeah, that's right, the whole eternity of our data center. Then you talk about our-- These frameworks, like Theano, Caffe, TensorFlow, right? These are new frameworks. They are machine learning and deep learning frameworks. They're open source and others. Then you go to libraries. Then you go to accelerators, which accelerators you choose, then you go to your operating systems. Now, you haven't even talked about your use case. Retail use case or genomic sequencing use case. All you're trying to do is now figure out TensorFlow works with this accelerator or does not work with this accelerator. Or, does Caffe and Theano work with this operating system or not? And, that is a complexity that is way more complex. So, that's where we felt that we really needed to launch these new solutions, and we prelaunched them here at Super Computing, because we feel the evolution of HBC towards AI is happening. We're going to start shipping these Ready Bundles for machine learning and deep learning in first half of 2018. >> So, that's what the Ready Solutions are? You're basically putting the solution together for the client, then they can start-- You work together to build the application to fix whatever it is they're trying to do. >> That's exactly it. But, not just fix it. It's an outcome. So, I'm going to go back to the retailer. So, if you are the CEO of the biggest retailer and you are saying, hey, I just don't want to know who buys from me, I want to now do predictive analytics, which is who buys chips and beer, but who can I sell more things to, right? So, you now start thinking about demographic data. You start thinking about payroll data and other datas that surround-- You start feeding that data into it, so your machine now starts to learn a lot more of those frameworks, and then can actually give you predictive analytics. But, imagine a day where you actually-- The machine or the deep learning AI actually tells you that it's not just who you want to sell chips and beer to, it's who's going to buy the 4k TV? You're makin' a lot of presumptions. Well, there you go, and the 4k-- But, I'm glad you're doin' the 4k TV. So, that's important, right? That is where our customers need to understand how predictive analytics are going to move towards cognitive analytics. So, this is complex but we're trying to make that complex simple with these Ready Solutions from machine learning and deep learning. >> So, I want to just get your take on-- You've kind of talked about these three things a couple times, how you delineate between AI, machine learning, and deep learning. >> So, as I said, there is an evolution. I don't think a customer can achieve artificial intelligence unless they go through the whole crawl walk around space. There's no shortcuts there, right? What do you do? So, if you think about, Mastercard is a great customer of ours. They do an incredible amount of transactions per day, (laughs) as you can think, right? In millions. They want to do facial recognitions at kiosks, or they're looking at different policies based on your buying behavior-- That, hey, Jeff doesn't buy $20,000 Rolexes every year. Maybe once every week, you know, (laughs) it just depends how your mood is. I was in the Emirates. Exactly, you were in Dubai (laughs). Then, you think about his credit card is being used where? And, based on your behaviors that's important. Now, think about, even for Mastercard, they have traditional RDBMS databases. They went to BI. They have high-performance computing clusters. Then, they developed the hadoop cluster. So, what we did with them, we said okay. All that is good. That data that has been generated for you through customers and through internal IT organizations, those things are all very important. But, at the same time, now you need to start going through this data and start analyzing this data for predictive analytics. So, they had 1.2 million policies, for example, that they had to crunch. Now, think about 1.2 million policies that they had to say-- In which they had to take decisions on. That they had to take decisions on. One of the policies could be, hey, does Jeff go to Dubai to buy a Rolex or not? Or, does Jeff do these other patterns, or is Armughan taking his card and having a field day with it? So, those are policies that they feed into machine learning frameworks, and then machine learning actually gives you patterns that they can now see what your behavior is. Then, based on that, eventually deep learning is when they move to next. Deep learning now not only you actually talk about your behavior patterns on the credit card, but your entire other life data starts to-- Starts to also come into that. Then, now, you're actually talking about something before, that's for catching a fraud, you can actually be a lot more predictive about it and cognitive about it. So, that's where we feel that our Ready Solutions around machine learning and deep learning are really geared towards, so taking HBC to then democratizing it, advancing it, and then now helping our customers move towards machine learning and deep learning, 'cause these buzzwords of AIs are out there. If you're a financial institution and you're trying to figure out, who is that customer who's going to buy the next mortgage from you? Or, who are you going to lend to next? You want the machine and others to tell you this, not to take over your life, but to actually help you make these decisions so that your bottom line can go up along with your top line. Revenue and margins are important to every customer. >> It's amazing on the credit card example, because people get so pissed if there's a false positive. With the amount of effort that they've put into keep you from making fraudulent transactions, and if your credit card ever gets denied, people go bananas, right? The behavior just is amazing. But, I want to ask you-- We're comin' to the end of 2017, which is hard to believe. Things are rolling at Dell EMC. Michael Dell, ever since he took that thing private, you could see the sparkle in his eye. We got him on a CUBE interview a few years back. A year from now, 2018. What are we going to talk about? What are your top priorities for 2018? >> So, number one, Michael continues to talk about that our vision is advancing human progress through technology, right? That's our vision. We want to get there. But, at the same time we know that we have to drive IT transformation, we have to drive workforce transformation, we have to drive digital transformation, and we have to drive security transformation. All those things are important because lots of customers-- I mean, Jeff, do you know like 75% of the S&P 500 companies will not exist by 2027 because they're either not going to be able to make that shift from Blockbuster to Netflix, or Uber taxi-- It's happened to our friends at GE over the last little while. >> You can think about any customer-- That's what Michael did. Michael actually disrupted Dell with Dell technologies and the acquisition of EMC and Pivotal and VMWare. In a year from now, our strategy is really about edge to core to the cloud. We think the world is going to be all three, because the rise of 20 billion devices at the edge is going to require new computational frameworks. But, at the same time, people are going to bring them into the core, and then cloud will still exist. But, a lot of times-- Let me ask you, if you were driving an autonomous vehicle, do you want that data-- I'm an Edge guy. I know where you're going with this. It's not going to go, right? You want it at the edge, because data gravity is important. That's where we're going, so it's going to be huge. We feel data gravity is going to be big. We think core is going to be big. We think cloud's going to be big. And we really want to play in all three of those areas. >> That's when the speed of light is just too damn slow, in the car example. You don't want to send it to the data center and back. You don't want to send it to the data center, you want those decisions to be made at the edge. Your manufacturing floor needs to make the decision at the edge as well. You don't want a lot of that data going back to the cloud. All right, Armughan, thanks for bringing the energy to wrap up our day, and it's great to see you as always. Always good to see you guys, thank you. >> All right, this is Armughan, I'm Jeff Frick. You're watching theCUBE from Super Computing Summit 2017. Thanks for watching. We'll see you next time. (soft electronic music)

>> Announcer: From Denver, Colorado it's theCUBE. Covering Super Computing 17, brought to you by Intel. Welcome back everybody, Jeff Frick here with theCUBE. We're at Super Computing 2017 in Denver, Colorado talking about big big iron, we're talking about space and new frontiers, black holes, mapping the brain. That's all fine and dandy, but we're going to have a little bit more fun this next segment. We're excited to have our next guest Bernie Spang. He's a VP Software Defined Infrastructure for IBM. And his buddy and guest Wayne Glanfield HPC Manager for Red Bull Racing. And for those of you that don't know, that's not the pickup trucks, it's not the guy jumping out of space, this is the Formula One racing team. The fastest, most advanced race cars in the world. So gentlemen, first off welcome. Thank you. Thank you Jeff. So what is a race car company doing here for a super computing conference? Obviously we're very interested in high performance computing so traditionally we've used a wind tunnel to do our external aerodynamics. HPC allows us to do many many more iterations, design iterations of the car. So we can actually kind of get more iterations of the designs out there and make the car go faster very quicker. So that's great, you're not limited to how many times you can get it in the wind tunnel. The time you have in the wind tunnel. I'm sure there's all types of restrictions, cost and otherwise. There's lots of restrictions and both the wind tunnel and in HPC usage. So with HPC we're limited to 25 teraflops, which isn't many teraflops. 25 teraflops. >> Wayne: That's all. And Bernie, how did IBM get involved in Formula One racing? Well I mean our spectrum computing offerings are about virtualizing clusters to optimize efficiency, and the performance of the workloads. So our Spectrum LSF offering is used by manufacturers, designers to get ultimate efficiency out of the infrastructure. So with the Formula One restrictions on the teraflops you want to get as much work through that system as efficiently as you can. And that's where Spectrum computing comes in. That's great. And so again, back to the simulations. So not only can you just do simulations 'cause you got the capacity, but then you can customize it as you said I think before we turned on the cameras for specific tracks, specific race conditions. All types of variables that you couldn't do very easily in a traditional wind tunnel. Yes obviously it takes a lot longer to actually kind of develop, create, and rapid prototype the models and get them in the wind tunnel, and actually test them. And it's obviously much more expensive. So by having a HPC facility we can actually kind of do the design simulations in a virtual environment. So what's been kind of the ahah from that? Is it just simply more better faster data? Is there some other kind of transformational thing that you observed as a team when you started doing this type of simulation versus just physical simulation in a wind tunnel? We started using HPC and computational fluid dynamics about 12 years ago in anger. Traditionally it started out as a complementary tool to the wind tunnel. But now with the advances in HPC technology and software from IBM, it's actually beginning to overtake the wind tunnel. So it's actually kind of driving the way we design the car these days. That's great. So Bernie, working with super high end performance, right, where everything is really optimized to get that car to go a little bit faster, just a little bit faster. Right. Pretty exciting space to work in, you know, there's a lot of other great applications, aerospace, genomics, this and that. But this is kind of a fun thing you can actually put your hands on. Oh it's definitely fun, it's definitely fun being with the Red Bull Racing team, and with our clients when we brief them there. But we have commercial clients in automotive design, aeronautics, semiconductor manufacturing, where getting every bit of efficiency and performance out of their infrastructure is also important. Maybe they're not limited by rules, but they're limited by money, you know and the ability to investment. And their ability to get more out of the environment gives them a competitive advantage as well. And really what's interesting about racing, and a lot of sports is you get to witness the competition. We don't get to witness the competition between big companies day to day. You're not kind of watching it in those little micro instances. So the good thing is you get to learn a lot from such a focused, relatively small team as Red Bull Racing that you can apply to other things. So what are some of the learnings as you've got work with them that you've taken back? Well certainly they push the performance of the environment, and they push us, which is a great thing for us, and for our other clients who benefit. But one of the things I think that really stands out is the culture there of the entire team no matter what their role and function. From the driver on down to everybody else are focused on winning races and winning championships. And that team view of getting every bit of performance out of everything everybody does all the time really opened our thinking to being broader than just the scheduling of the IT infrastructure, it's also about making the design team more productive and taking steps out of the process, and anything we can do there. Inclusive of the storage management, and the data management over time. So it's not just the compute environment it's also the virtualized storage environment. Right, and just massive amounts of storage. You said not only are you running and generating, I'm just going to use boatloads 'cause I'm not sure which version of the flops you're going to use. But also you got historical data, and you have result data, and you have models that need to be tweaked, and continually upgraded so that you do better the following race. Exactly, I mean we're generating petabytes of data a year and I think one of the issues which is probably different from most industries is our workflows are incredibly complex. So we have up to 200 discrete job steps for each workflow to actually kind of produce a simulation. This is where the kind of IBM Spectrum product range actually helps us do that efficiently. If you imagine an aerospace engineer, or aerodynamics engineer trying to manually manage 200 individual job steps, it just wouldn't happen very efficiently. So this is where Spectrum scale actually kind of helps us do that. So you mentioned it briefly Bernie, but just a little bit more specifically. What are some of the other industries that you guys are showcasing that are leveraging the power of Spectrum to basically win their races. Yeah so and we talked about the infrastructure and manufacturing, but they're industrial clients. But also in financial services. So think in terms of risk analytics and financial models being an important area. Also healthcare life sciences. So molecular biology, finding new drugs. When you talk about the competition and who wins right. Genomics research and advances there. Again, you need a system and an infrastructure that can chew through vast amounts of data. Both the performance and the compute, as well as the longterm management with cost efficiency of huge volumes of data. And then you need that virtualized cluster so that you can run multiple workloads many times with an infrastructure that's running in 80%, 90% efficiency. You can't afford to have silos of clusters. Right we're seeing clients that have problems where they don't have this cluster virtualization software, have cluster creep, just like in the early days we had server sprawl, right? With a different app on a different server, and we needed to virtualize the servers. Well now we're seeing cluster creep. Right the Hadoop clusters and Spark clusters, and machine learning and deep learning clusters. As well as the traditional HPC workload. So what Spectrum computing does is virtualizes that shared cluster environment so that you can run all these different kind of workloads and drive up the efficiency of the environment. 'Cause efficiency is really the key right. You got to have efficiency that's what, that's really where cloud got its start, you know, kind of eating into the traditional space, right. There's a lot of inefficient stuff out there so you got to use your resources efficiently it's way too competitive. Correct well we're also seeing inefficiencies in the use of cloud, right. >> Jeff: Absolutely. So one of the features that we've added to the Spectrum computing recently is automated dynamic cloud bursting. So we have clients who say that they've got their scientists or their design engineers spinning up clusters in the cloud to run workloads, and then leaving the servers running, and they're paying the bill. So we built in automation where we push the workload and the data over the cloud, start the servers, run the workload. When the workload's done, spin down the servers and bring the data back to the user. And it's very cost effective that way. It's pretty fun everyone talks often about the spin up, but they forget to talk about the spin down. Well that's where the cost savings is, exactly. Alright so final words, Wayne, you know as you look forward, it's super a lot of technology in Formula One racing. You know kind of what's next, where do you guys go next in terms of trying to get another edge in Formula One racing for Red Bull specifically. I mean I'm hoping they reduce the restrictions on HPC so it can actually start using CFD and the software IBM provides in a serious manner. So it can actually start pushing the technologies way beyond where they are at the moment. It's really interesting that they, that as a restriction right, you think of like plates and size of the engine, and these types of things as the rule restrictions. But they're actually restricting based on data size, your use of high performance computing. They're trying to save money basically, but. It's crazy. So whether it's a rule or you know you're share holders, everybody's trying to save money. Alright so Bernie what are you looking at, sort of 2017 is coming to an end, it's hard for me to say that as you look forward to 2018 what are some of your priorities for 2018. Well the really important thing and we're hearing it at this conference, I'm talking with the analysts and with the clients. The next generation of HPC in analytics is what we're calling machine learning, deep learning, cognitive AI, whatever you want to call it. That's just the new generation of this workload. And our Spectrum conductor offering and our new deep learning impact capability to automate the training of deep learning models, so that you can more quickly get to an accurate model like in hours or minutes, not days or weeks. That's going to a huge break through. And based on our early client experience this year, I think 2018 is going to be a breakout year for putting that to work in commercial enterprise use cases. Alright well I look forward to the briefing a year from now at Super Computing 2018. Absolutely. Alright Bernie, Wayne, thanks for taking a few minutes out of your day, appreciate it. You're welcome, thank you. Alright he's Bernie, he's Wayne, I'm Jeff Frick we're talking Formula One Red Bull Racing here at Super Computing 2017. Thanks for watching.

>> [Announcer] From Denver, Colorado, it's the Cube covering Super Computing 17, brought to you by Intel. (techno music) >> Welcome back, everybody, Jeff Frick with the Cube. We are at Super Computing 2017 here in Denver, Colorado. 12,000 people talking about big iron, heavy lifting, stars, future mapping the brain, all kinds of big applications. We're here, first time ever for the Cube, great to be here. We're excited for our next guest. She's Susan Bobholtz, she's the Fabric Alliance Manager for Omni-Path at Intel, Susan, welcome. >> Thank you. >> So what is Omni-Path, for those that don't know? >> Omni-Path is Intel's high performance fabric. What it does is it allows you to connect systems and make big huge supercomputers. >> Okay, so for the royal three-headed horsemen of compute, store, and networking, you're really into data center networking, connecting the compute and the store. >> Exactly, correct, yes. >> Okay. How long has this product been around? >> We started shipping 18 months ago. >> Oh, so pretty new? >> Very new. >> Great, okay and target market, I'm guessing has something to do with high performance computing. >> (laughing) Yes, our target market is high performance computing, but we're also seeing a lot of deployments in artificial intelligence now. >> Okay and so what's different? Why did Intel feel compelled that they needed to come out with a new connectivity solution? >> We were getting people telling us they were concerned that the existing solutions were becoming too expensive and weren't going to scale into the future, so they said Intel, can you do something about it, so we did. We made a couple of strategic acquisitions, we combined that with some of our own IP and came up with Omni-Path. Omni-Path is very much a proprietary protocol, but we use all the same software interfaces as InfiniBand, so your software applications just run. >> Okay, so to the machines it looks like InfiniBand? >> Yes. >> Just plug and play and run. >> Very much so, it's very similar. >> Okay what are some of the attributes that make it so special? >> The reason it's really going very well is that it's the price performance benefits, so we have equal to, or better, performance than InfiniBand today, but we also have our switch technology is 48 ports verses InfiniBand is 36 ports. So that means you can build denser clusters in less space and less cables, lower power, total cost of ownership goes down, and that's why people are buying it. >> Really fits into the data center strategy that Intel's executing very aggressively right now. >> Fits very nicely, absolutely, yes, very much so. >> Okay, awesome, so what are your thoughts here at the show? Any announcements, anything that you've seen that's of interest? >> Oh yeah, so, a couple things. We've had really had good luck on the Top 500 list. 60% of the servers that are running a 100 gigabyte fabrics in the Top 500 list are running connected via Omni-Path. >> What percentage again? >> 60% >> 60? >> Yes. >> You've only been at it for 18 months? >> Yes, exactly. >> Impressive. >> Very, very good. We've got systems in the Top 10 already. Some of the Top 10 systems in the world are using Omni-Path. >> Is it rip and replace, do you find, or these are new systems that people are putting in. >> Yeah, these are new systems. Usually when somebody's got a system they like and run, they don't want to touch it. >> Right. >> These are people saying I need a new system. I need more power, I need more oompf. They have the money, the budget, they want to put in something new, and that's when they look to Omni-Path. >> Okay, so what are you working on now, what's kind of next for Omni-Path? >> What's next for us is we are announcing a new higher, denser switch technology, so that will allow you to go for your director class switches, which is the really big ones, is now rather than having 768 ports, you can go to 1152, and that means, again, denser topologies, lower power, less cabling, it reduces your total cost of ownership. >> Right, I think you just answered my question, but I'm going to ask you anyway. >> (laughs) Okay. >> We talked a little bit before we turned the camera on about AI and some of the really unique challenges of AI, and that was part of the motivation behind this product. So what are some of the special attributes of AI that really require this type of connectivity? >> It's very much what you see even with high performance computing. You need low latency, you need high bandwidth. It's the same technologies, and in fact, in a lot of cases, it's the same systems, or sometimes they can be running software load that is HPC focused, and sometimes they're running a software load that is artificial intelligence focused. But they have the same exact needs. >> Okay. >> Do it fast, do it quick. >> Right, right, that's why I said you already answered the question. Higher density, more computing, more storing, faster. >> Exactly, right, exactly. >> And price performance. All right, good, so if we come back a year from now for Super Computing 2018, which I guess is in Dallas in November, they just announced. What are we going to be talking about, what are some of your priorities and the team's priorities as you look ahead to 2018? >> Oh we're continuing to advance the Omni-Path technology with software and additional capabilities moving forward, so we're hoping to have some really cool announcements next year. >> All right, well, we'll look forward to it, and we'll see you in Dallas in a year. >> Thanks, Cube. >> All right, she's Susan, and I'm Jeff. You're watching the Cube from Super Computing 2017. Thanks for watching, see ya next time. (techno music)

>> Announcer: From Denver, Colorado, it's theCUBE covering Super Computing '17. Brought to you by Intel. (upbeat techno music) Hey welcome back, everybody. Jeff Frick here with theCUBE. We're at Super Computing 2017 in Denver, Colorado. It's our first trip to the show, 12,000 people, a lot of exciting stuff going on, big iron, big lifting, heavy duty compute. We're excited to have our next guest on. He's Jim Wu, he's the Director of Customer Experience for Falcon Computing. Jim, welcome. Thank you. Good to see you. So, what does Falcon do for people that aren't familiar with the company? Yeah, Falcon Company is in our early stages startup, focus on AVA-based acceleration development. Our vision is to allow software engineers to develop a FPGA-based accelerators, accelerators without FPGA expertise. Right, you just said you closed your B round. So, congratulations on that. >> Jim: Thank you. Yeah, very exciting. So, it's a pretty interesting concept. To really bring the capability to traditional software engineers to program for hardware. That's kind of a new concept. What do you think? 'Cause it brings the power of a hardware system. but the flexibility of a software system. Yeah, so today, to develop FPGA accelerators is very challenging. So, today for the accelerations-based people use very low level language, like a Verilog and the VHDL to develop FPGA accelerators. Which was very time consuming, very labor-intensive. So, our goal is to liberate them to use, C/C++ space design flow to give them an environment that they are familiar with in C/C++. So now not only can they improve their productivity, we also do a lot of automatic organization under the hood, to give them the highest accelerator results. Right, so that really opens up the ecosystem well beyond the relatively small ecosystem that knows how to program their hardware. Definitely, that's what we are hoping to see. We want to the tool in the hands of all software programmers. They can use it in the Cloud. They can use it on premises. Okay. So what's the name of your product? And how does it fit within the stack? I know we've got the Intel microprocessor under the covers, we've got the accelerator, we've got the cards. There's a lot of pieces to the puzzle. >> Jim: Yeah. So where does Falcon fit? So our main product is a compiler, called the Merlin Compiler. >> Jeff: Okay. It's a pure C and the C++ flow that enables software programmers to design APGA-based accelerators without any knowledge of APGA. And it's highly integrated with Intel development tools. So users don't even need to learn anything about the Intel development environment. They can just use their C++ development environment. Then in the end, we give them the host code as well as APGA binaries so they can round on APGA to see a accelerated applications. Okay, and how long has Merlin been GA? Actually, we'll be GA early next year. Early next year. So finishing, doing the final polish here and there. Yes. So in this quarter, we are heavily investing a lot of ease-of-use features. Okay. We have most of the features we want to be in the tool, but we're still lacking a bit in terms of ease-of-use. >> Jeff: Okay. So we are enhancing our report capabilities, we are enhancing our profiling of capabilities. We want to really truly like a traditional C++-based development environment for software application engineers. Okay, that's fine. You want to get it done, right, before you ship it out the door? So you have some Alpha programs going on? Some Beta programs of some really early adopters? Yeah, exactly. So today we provide a 14 day free trial to any customers who are interested. We have it, you can set up your enterprise or you can set up on Cloud. Okay. We provide to where you want your work done. Okay. And so you'll support all the cloud service providers, the big public clouds, all the private clouds. All the traditional data servers as well. Right. So, we are twice already on Aduplas as well as Alibaba Cloud. So we are working on bringing the tool to other public cloud providers as well. Right. So what is some of the early feedback you're getting from some of the people you're talking to? As to where this is going to make the biggest impact. What type of application space has just been waiting for this solution? So our Merlin Compiler is a productivity tool, so any space that FPGA can traditionally play well that's where we want to be there. So like encryption, decryption, video codec, compression, decompression. Those kind of applications are very stable for APGA. Now traditionally they can only be developed by hardware engineers. Now with the Merlin Compiler, all of these software engineers can use the Merlin Compiler to do all of these applications. Okay. And when is the GA getting out, I know it's coming? When is it coming? Approximately So probably first quarter of 2018. Okay, that's just right around the corner. Exactly. Alright, super. And again, a little bit about the company, how many people are you? A little bit of the background on the founders. So we have about 30 employees, at the moment, so we have offices in Santa Clara which is our headquarters. We also have an office in Los Angeles. As well as a Beijing, China. Okay, great. Alright well Jim, thanks for taking a few minutes. We'll be looking for GA in a couple of months and wish you nothing but the best success. Okay, thank you so much, Jeff. Alright, he's Jim Lu I'm Jeff Frick. You're watching theCUBE from supering computing 2017. Thanks for watching. (upbeat techno music)

>> Narrator: From Denver, Colorado, it's theCUBE. Covering Super Computing '17, brought to you by Intel. (electronic music) >> Hey, welcome back everybody. Jeff Frick here with theCUBE. We're at Denver, Colorado at Super Computing 2017. 12,000 people, our first trip to the show. We've been trying to come for awhile, it's pretty amazing. A lot of heavy science in terms of the keynotes. All about space and looking into brain mapping and it's heavy lifting, academics all around. We're excited to have our next guest, who's an expert, all about speed and that's John Lockwood. He's the CEO of Algo-Logic. First off, John, great to see you. >> Yeah, thanks Jeff, glad to be here. >> Absolutely, so for folks that aren't familiar with the company, give them kind of the quick overview of Algo. >> Yes, Algo-Logic puts algorithms into logic. So our main focus is taking things are typically done in software and putting them into FPGAs and by doing that we make them go faster. >> So it's a pretty interesting phenomenon. We've heard a lot from some of the Intel execs about kind of the software overlay that now, kind of I guess, a broader ecosystem of programmers into hardware, but then still leveraging the speed that you get in hardware. So it's a pretty interesting combination to get those latencies down, down, down. >> Right, right, I mean Intel certainly made a shift to go on into heterogeneous compute. And so in this heterogeneous world, we've got software running on Xeons, Xeon Phis. And we've also got the need though, to use new compute in more than just the traditional microprocessor. And so with the acquisition of Altera, is that now Intel customers can use FPGAs in order to get the benefit in speed. And so Algo-Logic, we typically provide applications with software APIs, so it makes it really easy for end customers to deploy FPGAs into their data center, into their hosts, into their network and start using them right away. >> And you said one of your big customer sets is financial services and trading desk. So low latency there is critical as millions and millions and millions if not billions of dollars. >> Right, so Algo-Logic we have a whole product line of high-frequency trading systems. And so our Tick-To-Trade system is unique in the fact that it has a sub-microsecond trading latency and this means going from market data that comes in, for example on CME for options and futures trading, to time that we can place a fix order back out to the market. All of that happens in an FPGA. That happens in under a microsecond. So under a millionth of second and that beats every other software system that's being used. >> Right, which is a game change, right? Wins or losses can be made on those time frames. >> It's become a must have is that if you're trading on Wall Street or trading in Chicago and you're not trading with an FPGA, you're trading at a severe disadvantage. And so we make a product that enables all the trading firms to be playing on a fair, level playing field against the big firms. >> Right, so it's interesting because the adoption of Flash and some of these other kind of speed accelerator technologies that have been happening over the last several years, people are kind of getting accustomed to the fact that speed is better, but often it was kind of put aside in this kind of high-value applications like financial services and not really proliferating to a broader use of applications. I wonder if you're seeing that kind of change a little bit, where people are seeing the benefits of real time and speed beyond kind of the classic high-value applications? >> Well, I think the big change that's happened is that it's become machine-to-machine now. And so humans, for example in trading, are not part of the loop anymore and so it's not a matter of am I faster than another person? It's am I faster than the other person's machine? And so this notion of having compute that goes fast has become suddenly dramatically much more important because everything now is going to machine versus machine. And so if you're an ad tech advertiser, is that how quickly you can do an auction to place an ad matters and if you can get a higher value ad placed because you're able to do a couple rounds of an auction, that's worth a lot. And so, again, with Algo-Logic we make things go faster and that time benefit means, that all thing else being the same, you're the first to come to a decision. >> Right, right and then of course the machine-to-machine obviously brings up the hottest topic that everybody loves to talk about is autonomous vehicles and networked autonomous vehicles and just the whole IOT space with the compute moving out to the edge. So this machine-to-machine systems are only growing in importance and really percentage of the total compute consumption by far. >> That's right, yeah. So last year at Super Computing, we demonstrated a drone, bringing in realtime data from a drone. So doing realtime data collection and doing processing with our Key Value Store. So this year, we have a machine learning application, a Markov Decision Process where we show that we can scale-out a machine learning process and teach cars how to drive in a few minutes. >> Teach them how to drive in a few minutes? >> Right. >> So that's their learning. That's not somebody programming the commands. They're actually going through a process of learning? >> Right, well so the Key Value Store is just a part of this. We're just the part of the system that makes the scale-outs that runs well in a data center. And so we're still running the Markov Decision Process in simulations in software. So we have a couple Xeon servers that we brought with us to do the machine learning and a data center would scale-out to be dozens of racks, but even with a few machines though, for simple highway driving, what we can show is we start off with, the system's untrained and that in the Markov Decision Process, we reward the final state of not having accidents. And so at first, the cars drive and they're bouncing into each other. It's like bumper cars, but within a few minutes and after about 15 million simulations, which can be run that quickly, is that the cars start driving better than humans. And so I think that's a really phenomenal step, is the fact that you're able to get to a point where you can train a system how to drive and give them 15 man years of experience in a matter of minutes by the scale-out compute systems. >> Right, 'cause then you can put in new variables, right? You can change that training and modify it over time as conditions change, throw in snow or throw in urban environments and other things. >> Absolutely, right. And so we're not pretending that our machine learning, that application we're showing here is an end-all solution. But as you bring in other factors like pedestrians, deer, other cars running different algorithms or crazy drivers, is that you want to expose the system to those conditions as well. And so one of the questions that came up to us was, "What machine learning application are you running?" So we're showing all 25 cars running one machine learned application and that's incrementally getting better as they learn to drive, but we could also have every car running a different machine learning application and see how different AIs interact with each other. And I think that's what you're going to see on the highway as we have more self-driving cars running different algorithms, we have to make sure they all place nice with each other. >> Right, but it's really a different way of looking at the world, right, using machine learning, machine-to-machine versus single person or a team of people writing a piece of software to instruct something to do something and then you got to go back and change it. This is a much more dynamic realtime environment that we're entering into with IOT. >> Right, I mean the machine-to-human, which was kind of last year and years before, were, "How do you make interactions "between the computers better than humans?" But now it's about machine-to-machine and it's,"How do you make machines interact better "with other machines?" And that's where it gets really competitive. I mean, you can imagine with drones for example, for applications where you have drones against drones, the drones that are faster are going to be the ones that win. >> Right, right, it's funny, we were just here last week at the commercial drone show and it's pretty interesting how they're designing the drones now into a three-part platform. So there's the platform that flies around. There's the payload, which can be different sensors or whatever it's carrying, could be herbicide if it's an agricultural drone. And then they've opened up the STKs, both on the control side as well as the mobile side, in terms of the controls. So it's a very interesting way that all these things now, via software could tie together, but as you say, using machine learning you can train them to work together even better, quicker, faster. >> Right, I mean having a swarm or a cluster of these machines that work with each other, you could really do interesting things. >> Yeah, that's the whole next thing, right? Instead of one-to-one it's many-to-many. >> And then when swarms interact with other swarms, then I think that's really fascinating. >> So alright, is that what we're going to be talking about? So if we connect in 2018, what are we going to be talking about? The year's almost over. What are your top priorities for next year? >> Our top priorities are to see. We think that FPGA is just playing this important part. A GPU for example, became a very big part of the super computing systems here at this conference. But the other side of heterogeneous is the FPGA and the FPGA has seen almost, just very minimal adoption so far. But the FPGA has the capability of providing, especially when it comes to doing network IO transactions, it's speeding up realtime interactions, it has an ability to change the world again for HPC. And so I'm expecting that in a couple years, at this HPC conference, that what we'll be talking about, is the biggest top 500 super computers, is that how big of FPGAs do they have. Not how big of GPUs do they have. >> All right, time will tell. Well, John, thanks for taking a few minutes out of your day and stopping by. >> Okay, thanks Jeff, great to talk to you. >> All right, he's John Lockwood, I'm Jeff Frick. You're watching theCUBE from Super Computing 2017. Thanks for watching. >> Bye. (electronic music)

>> Voiceover: From Denver, Colorado, it's theCUBE covering Super Computing '17, brought to you by Intel. Hey, welcome back, everybody. Jeff Frick, here, with theCUBE. We're in Denver, Colorado at Super Computing 2017. It's all things heavy lifting, big iron, 12,000 people. I think it's the 20th anniversary of the conference. A lot of academics, really talking about big iron, doin' big computing. And we're excited to have our next guest, talking about speed, he's Stephane Monoboisset. Did I get that right? That's right. He's a director of marketing and partnerships for Accelize. Welcome. Thank you. So, for folks that aren't familiar with Accelize, give them kind of the quick overview. Okay, so Accelize is a French startup. Actually, a spinoff for a company called PLDA that has been around for 20 years, doing PCI express IP. And about a few years ago, we started initiative to basically bring FPGA acceleration to the cloud industry. So what we say is, we basically enable FPGA acceleration as a service. So did it not exist in cloud service providers before that, or what was kind of the opportunity that you saw there? So, FPGAs have been used in data centers in many different ways. They're starting to make their way into, as a service type of approach. But one of the thing that the industry, one of the buzzword that the industry's using, is FPGA as a service. And the industry usually refers to it as the way to bring FPGA to the end users. But when you think about it, end users don't really want FPGA as a service. Most of the cloud end users are not FPGA experts. So they couldn't care less whether it's an FPGA or something else. What they really want is the acceleration benefits. Hence the term, FPGA acceleration as a service. So, in order to do that, instead of just going and offering an FPGA platform, and giving them the tools, even if they are easy to use and develop the FPGAs, our objective is to propose to provide a marketplace of accelerators that they can use as a service, without even thinking that it's an FPGA on the background. So that's a really interesting concept. Because that also leverages an ecosystem. And one thing we know that's important, if you have any kind of a platform playing, you need an ecosystem that brings a much broader breadth of applications, and solution suites, and there's a lot of talk about solutions. So that was pretty insightful, 'cause now you open it up to this much broader set of applications. Well, absolutely. The ecosystem is the essential part of the offering because obviously, as a company, we cannot be expert in every single domain. And to a certain extent, even FPGA designers, they are what, about maybe 10, 15,000 FPGA designers in the world. They are not really expert in the end application. So one of the challenges that we're trying to address is how do we make application developers, the people who are already playing in the cloud, the ISVs, for example, who have the expertise of what the end user wants, being able to develop something that is efficient to the end user in FPGAs. And this is why we've created a tool called Quick Play, which basically enables what we call the accelerator function developers, the guys who have the application expertise, to leverage an ecosystem of IP providers in the FPGA space that have built efficient building blocks, like encryption, compression, video transcoding. Right. These sort of things. So what you have is an ecosystem of cloud service providers. You have an ecosystem of IP providers. And we have this growing ecosystem of accelerator developers that develop all these accelerators that are sold as a service. And that really opens up the number of people that are qualified to play in the space. 'Cause you're kind of hiding the complexity into the hardcore, harder engineers and really making it more kind of a traditional software application space. Is that right? Yeah, you're absolutely right. And we're doing that on the technical front, but we're also doing that on the business model front. Because one thing with FPGAs is that FPGAs has relied heavily over the years on the IP industry. And the IP industry for FPGAs, and it's the same for ASIGs, have been also relying on the business model, which is based on very high up-front cost. So let me give you an example. Let's say I want to develop an accelerator, right, for database. And what I need to do is to get the stream of data coming in. It's most likely encrypted, so I need to decrypt this data, then I want to do some search algorithm on it to extract certain functions. I'm going to do some processing on it, and maybe the last thing I want to do is, I want to compress because I want to store the result of that data. If I'm doing that with a traditional IP business model, what I need to do is basically go and talk to every single one of those IP providers and ask them to sell me the IP. In the traditional IP business model, I'm looking at somewhere between 200,000 to 500,000 up front cost. And I want to sell this accelerator for maybe a couple of dollars on one of the marketplace. There's something that doesn't play out. So what we've done, also, is we've introduced a pay-per-use business model that allows us to track those IPs that are being used by the accelerators so we can propagate the as-a-service business model throughout the industry, the supply chain. Which is huge, right? 'Cause as much as cloud is about flexibility and extensibility, it's about the business model as well. About paying what you use when you use it, turning it on, turning it off. So that's a pretty critical success factor. Absolutely, I mean, you can imagine that there's, I don't know, millions of users in the cloud. There's maybe hundreds of thousands of different type of ways they're processing their data. So we also need a very agile ecosystem that can develop very quickly. And we also need them to do it in a way that doesn't cost too much money, right? Think about it, and think about the app store when it was launched, right? Right. When Apple launched the iPhone back about 10 years ago, right, they didn't have much application. And they didn't, I don't think they quite knew, exactly, how it was going to be used. But what they did, which completely changed the industry, is they opened up the SDK that they sold for very small amount of money and enabled a huge community to come up with a lot of a lot of application. And now you go there and you can find application that really meats your need. That's kind of the similar concept that we're trying to develop here. Right. So how's been the uptake? I mean, so where are you, kind of, in the life cycle of this project? 'Cause it's a relatively new spinout of the larger company? Yes, so it's relatively new. We did the spinout because we really want to give that product its own life. Right, right. Right? But we are still at the beginning. So we started a developing partnership with cloud service providers. The two ones that we've announced is Amazon Web Services and OVH, the cloud service provider in France. And we have recruited, I think, about a dozen IP partners. And now we're also working with accelerator developer, accelerator functions developers. Okay. So it's a work in progress. And our main goal right now is to, really to evangelize, and to show them how much money they can do and how they can serve this market of FPGA acceleration as a service. The cloud providers, or the application providers? Who do you really have to convince the most? So the one we have to convince today are really the application developers. Okay, okay. Because without content, your marketplace doesn't mean much. So this is the main thing we're focusing on right now. Okay, great. So, 2017's coming to an end, which is hard to believe. So as you look forward to 2018, of those things you just outlined, kind of what are some of the top priorities for 2018? So, top priorities will be to strengthen our relationship with the key cloud service providers we work with. We have a couple of other discussions ongoing to try to offer a platform on more cloud service providers. We also want to strengthen our relationship with Intel. And we'll continue the evangelization to really onboard all the IP providers and the accelerator developers so that the marketplace becomes filled with valuable accelerators that people can use. And that's going to be a long process, but we are focusing right now on key application space that we know people can leverage in the application. Exciting times. Oh yeah, it is. You know, it's 10 years since the app store launched, I think, so I look at acceleration as a service in cloud service providers, this sounds like a terrific opportunity. It is, it is a huge opportunity. Everybody's talking about it. We just need to materialize it now. All right, well, congratulations and thanks for taking a couple minutes out of your day. Oh, thanks for your time. All right, he's Stephane, I'm Jeff Frick. You're watching theCUBE from Super Computing 2017. Thanks for watching. (upbeat music)

>> Announcer: On Denver, Colorado, it's theCUBE, covering SuperComputing '17, brought to you by Intel. >> Hey, welcome back, everybody. Jeff Frick here with theCUBE. We're in Denver, Colorado at this SuperComputing conference 2017. I think there's 12,000 people. Our first time being here is pretty amazing. A lot of academics, a lot of conversations about space and genomes and you know, heavy-lifting computing stuff. It's fun to be here, and we're really excited. Our next guest, Karsten Ronner. He's the CEO of Swarm64. So Karsten, great to see you. >> Yeah, thank you very much for this opportunity. >> Absolutely. So for people that aren't familiar with Swarm64, give us kind of the quick eye-level. >> Yeah. Well, in a nutshell, Swarm64 is accelerating relational databases, and we allow them to ingest data so much faster, 50 times faster than a relational database. And we can also then query that data 10, 20 times faster than relational database. And that is very important for many new applications in IoT and in netbanking and in finance, and so on. >> So you're in a good space. So beyond just general or better performance, faster, faster, faster, you know, we're seeing all these movements now in real-time analytics and real-time applications, which is only going to get crazier with IoT and Internet of Things. So how do you do this? Where do you do this? What are some of the examples you could share with us? >> Yeah, so all our solution is a combination of a software wrapper that attaches our solution to existing databases. And inside, there's an FPGA from Intel, the Arria 10. And we are combining both, such that they actually plug into standard interfaces of existing databases, like in PostgreSQL, Foreign Data Wrappers, the storage engine in MySQL, and MariaDB and so on. And with that mechanism, we ensure that the database, the application doesn't see us. For the application, there's just fast database but we're invisible and also the functionality of the database remains what it was. That's the net of what we're doing. >> So that's so important because we talked a little bit about offline, you said you had a banking customer that said they have every database that's ever been created. They've been buying them all along so they've got embedded systems, you can't just rip and replace. You have to work with existing infrastructure. At the same time, they want to go faster. >> Yeah, absolutely right. Absolutely right. And there's a huge code base, which has been verified, which has been debugged, and in banking, it's also about compliance so you can't just rip out your old code base and do something new, because again, you would have to go through compliance. Therefore, customers really, really, really want their existing databases faster. >> Right. Now the other interesting part, and we've talked to some of the other Intel execs, is kind of this combination hybrid of the Hardware Software Solution in the FPGA, and you're really opening up an ecosystem for people to build more software-based solutions that leverage that combination of the hardware software power. Where do you see that kind of evolving? How's that going to help your company? >> Yeah. We are a little bit unique in that we are hiding that FPGA from the user, and we're not exposing it. Many people, actually, many applications expose it to the user, but apart from that, we are benefiting a lot from what Intel is doing. Intel is providing the entire environment, including virtualization, all those things that help us then to be able to get into Cloud service providers or into proprietary virtualized environments and things like that. So it is really a very close cooperation with Intel that helps us and enables us to do what we're doing. >> Okay. And I'm curious because you spend a lot of time with customers, you said a lot of legacy customers. So as they see the challenges of this new real-time environment, what are some of their concerns, what are some of the things that they're excited that they can do now with real-time, versus bash and data lake. And I think it's always funny, right? We used to make decisions based on stuff that happened in the past. And we're kind of querying now really the desires just to make action on stuff that's happening now, it's a fundamentally different way to address a problem. >> Yeah, absolutely. And a very, very key element of our solution is that we can not only insert these very, very large amounts of data that also other solutions can do, massively parallel solutions, streaming solutions, you know them all. They can do that too. However, the difference is that we can make that data available within less than 10 microseconds. >> Jeff: 10 microseconds? >> So dataset arrives within less than 10 microseconds, that dataset is part of the next query and that is a game changer. That allows you to do controlled loop processing of data in machine-to-machine environments, and autonomous, for autonomous applications and all those solutions where you just can't wait. If your car is driving down the street, you better know what has happened, right? And you can react to it. As an example, it could be a robot in a plant or things like that, where you really want to react immediately. >> I'm curious as to the kind of value unlocking that that provides to those old applications that were working with what they think is an old database. Now, you said, you know, you're accelerating it. To the application, it looks just the same as it looked before. How does that change those performances of those applications? I would imagine there's a whole other layer of value unlocking in those entrenched applications with this vast data. >> Yeah. That is actually true, and it's on a business level, the applications enable customers to do things they were not capable of doing before, and look for example in finance. If you can analyze the market data much quicker, if you can analyze past trades much quicker, then obviously you're generating value for the firm because you can react to market trends more accurately, you can mirror them in a more tighter fashion, and if you can do that, then you can reduce the margin of error with which you're estimating what's happening, and all of that is money. It's really pure money in the bank account of the customer, so to speak. >> Right. And the other big trend we talked about, besides faster, is you know, sampling versus not sampling. In the old days, we sampled old data and made decisions. Now we don't want to sample, we want all of the data, we want to make decisions on all the data, so again that's opening up another level of application performance because it's all the data, not a sample. >> For sure. Because before, you were aggregating. When you aggregate, you reduce the amount of information available. Now, of course, when you have the full set of information available, your decision-making is just so much smarter. And that's what we're enabling. >> And it's funny because in finance, you mentioned a couple of times, they've been doing that forever, right. The value of a few units of time, however small, is tremendous, but now we're seeing it in other industries as well that realize the value of real-time, aggregated, streaming data versus a sampling of old. Really opens up new types of opportunities. >> Absolutely, yes, yes. Yeah, finance, as I mentioned is an example, but then also IoT, machine-to-machine communication, everything which is real-time, logging, data logging, security and network monitoring. If you want to really understand what's flowing through your network, is there anything malicious, is there any actor on my network that should not be there? And you want to react so quickly that you can prevent that bad actor from doing anything to your data, this is where we come in. >> Right. And security's so big, right? It in everywhere. Especially with IoT and machine learning. >> Absolutely. >> All right, Karsten, I'm going to put you on the spot. So we're November 2017, hard to believe. As you look forward to 2018, what are some of your priorities? If we're standing here next year, at SuperComputing 2018, what are we going to be talking about? >> Okay, what we're going to talk about really is that we will, right now we're accelerating single-server solutions and we are working very, very hard on massively parallel systems, while retaining the real-time components. So we will not only then accelerate a single server, by then, allowing horizontal scaling, we will then bring a completely new level of analytics performance to customers. So that's what I'm happy to talk to you about next year. >> All right, we'll see you next year, I think it's in Texas. >> Wonderful, yeah, great. >> So thanks for stopping by. >> Thank you. >> He's Karsten, I'm Jeff. You're watching TheCUBE, from SuperComputing 2017. Thanks for watching.

>> Narrator: From Denver, Colorado, it's theCUBE. Covering Super Computing 17. Brought to you by Intel. (techno music) Hey, welcome back, everybody. Jeff Frick here with theCUBE. We're in Denver, Colorado at the Super Computing Conference 2017. About 12 thousand people, talking about the outer edges of computing. It's pretty amazing. The keynote was huge. The square kilometer array, a new vocabulary word I learned today. It's pretty exciting times, and we're excited to have our next guest. He's Bill Jenkins. He's a Product Line Manager for AI on FPGAs at Intel. Bill, welcome. Thank you very much for having me. Nice to meet you, and nice to talk to you today. So you're right in the middle of this machine-learning AI storm, which we keep hearing more and more about. Kind of the next generation of big data, if you will. That's right. It's the most dynamic industry I've seen since the telecom industry back in the 90s. It's evolving every day, every month. Intel's been making some announcements. Using this combination of software programming and FPGAs on the acceleration stack to get more performance out of the data center. Did I get that right? Sure, yeah, yeah. Pretty exciting. The use of both hardware, as well as software on top of it, to open up the solution stack, open up the ecosystem. What of those things are you working on specifically? I really build first the enabling technology that brings the FPGA into that Intel ecosystem. Where Intel is trying to provide that solution from top to bottom to deliver AI products. >> Jeff: Right. Into that market. FPGAs are a key piece of that because we provide a different way to accelerate those machine-learning and AI workloads. Where we can be an offload engine to a CPU. We can be inline analytics to offload the system, and get higher performance that way. We tie into that overall Intel ecosystem of tools and products. Right. So that's a pretty interesting piece because the real-time streaming data is all the rage now, right? Not in batch. You want to get it now. So how do you get it in? How do you get it written to the database? How do you get it into the micro-processor? That's a really, really important piece. That's different than even two years ago. You didn't really hear much about real-time. I think it's, like I said, it's evolving quite a bit. Now, a lot of people deal with training. It's the science behind it. The data scientists work to figure out what topologies they want to deploy and how they want to deploy 'em. But now, people are building products around it. >> Jeff: Right. And once they start deploying these technologies into products, they realize that they don't want to compensate for limitations in hardware. They want to work around them. A lot of this evolution that we're building is to try to find ways to more efficiently do that compute. What we call inferencing, the actual deployed machine-learning scoring, as they will. >> Jeff: Right. In a product, it's all about how quickly can I get the data out. It's not about waiting two seconds to start the processing. You know, in an autonomous-driven car where someone's crossing the road, I'm not waiting two seconds to figure out it's a person. Right, right. I need it right away. So I need to be able to do that with video feeds, right off a disk drive, from the ethernet data coming in. I want to do that directly in line, so that my processor can do what it's good at, and we offload that processor to get better system performance. Right. And then on the machine-learning specifically, 'cause that is all the rage. And it is learning. So there is a real-time aspect to it. You talked about autonomous vehicles. But there's also continuous learning over time, that's not necessarily dependent on learning immediately. Right. But continuous improvement over time. What are some of the unique challenges in machine-learning? And what are some of the ways that you guys are trying to address those? Once you've trained the network, people always have to go back and retrain. They say okay, I've got a good accuracy, but I want better performance. Then they start lowering the precision, and they say well, today we're at 32-bit, maybe 16-bit. Then they start looking into eight. But the problem is, their accuracy drops. So they retrain that into eight topology, that network, to get the performance benefit, but with the higher accuracy. The flexibility of FPGA actually allows people to take that network at 32-bit, with the 32-bit trained weights, but deploy it in lower precision. So we can abstract away the fact that the hardware's so flexible, we can do what we call floating point 11-bit floating point. Or even 8-bit floating point. Even here today at the show, we've got a binary and ternary demo, showcasing the flexibility that the FPGA can provide today with that building block piece of hardware that the FPGA can be. And really provide, not only the topologies that people are trying to build today, but tomorrow. >> Jeff: Right. Future proofing their hardware. But then the precisions that they may want to do. So that they don't have to retrain. They can get less than a 1% accuracy loss, but they can lower that precision to get all the performance benefits of that data scientist's work to come up with a new architecture. Right. But it's interesting 'cause there's trade-offs, right? >> Bill: Sure. There's no optimum solution. It's optimum as to what you're trying to optimize for. >> Bill: Right. So really, the ability to change the ability to continue to work on those learning algorithms, to be able to change your priority, is pretty key. Yeah, a lot of times today, you want this. So this has been the mantra of the FPGA for 30 plus years. You deploy it today, and it works fine. Maybe you build an ASIC out of it. But what you want tomorrow is going to be different. So maybe if it's changing so rapidly, you build the ASIC because there's runway to that. But if there isn't, you may just say, I have the FPGA, I can just reprogram it to do what's the next architecture, the next methodology. Right. So it gives you that future proofing. That capability to sustain different topologies. Different architectures, different precisions. To kind of keep people going with the same piece of hardware. Without having to say, spin up a new ASIC every year. >> Jeff: Right, right. Which, even then, it's so dynamic it's probably faster then, every year, the way things are going today. So the other thing you mentioned is topography, and it's not the same topography you mentioned, but this whole idea of edge. Sure. So moving more and more compute, and store, and smarts to the edge. 'Cause there's just not going to be time, you mentioned autonomous vehicles, a lot of applications to get everything back up into the cloud. Back into the data center. You guys are pushing this technology, not only in the data center, but progressively closer and closer to the edge. Absolutely. The data center has a need. It's always going to be there, but they're getting big. The amount of data that we're trying to process every day is growing. I always say that the telecom industry started the Information Age. Well, the Information Age has done a great job of collecting a lot of data. We have to process that. If you think about where, maybe I'll allude back to autonomous vehicles. You're talking about thousands of gigabytes, per day, of data generated. Smart factories. Exabytes of data generated a day. What are you going to do with all that? It has to be processed. We need that compute in the data center. But we have to start pushing it out into the edge, where I start thinking, well even a show like this, I want security. So, I want to do real-time weapons detection, right? Security prevention. I want to do smart city applications. Just monitoring how traffic moves through a mall, so that I can control lighting and heating. All of these things at the edge, in the camera, that's deployed on the street. In the camera that's deployed in a mall. All of that, we want to make those smarter, so that we can do more compute. To offload the amount of data that needs to be sent back to the data center. >> Jeff: Right. As much as possible. Relevant data gets sent back. No shortage of demand for compute store networking, is there? No, no. It's really a heterogeneous world, right? We need all the different compute. We need all the different aspects of transmission of the data with 5G. We need disk space to store it. >> Jeff: Right. We need cooling to cool it. It's really becoming a heterogeneous world. All right, well, I'm going to give you the last word. I can't believe we're in November of 2017. Yeah. Which is bananas. What are you working on for 2018? What are some of your priorities? If we talk a year from now, what are we going to be talking about? Intel's acquired a lot of companies over the past couple years now on AI. You're seeing a lot of merging of the FPGA into that ecosystem. We've got the Nervana. We've got Movidius. We've got Mobileye acquisitions. Saffron Technologies. All of these things, when the FPGA is kind of a key piece of that because it gives you that flexibility of the hardware, to extend those pieces. You're going to see a lot more stuff in the cloud. A lot more stuff with partners next year. And really enabling that edge to data center compute, with things like binary neural networks, ternary neural networks. All the different next generation of topologies to kind of keep that leading edge flexibility that the FPGA can provide for people's products tomorrow. >> Jeff: Exciting times. Yeah, great. All right, Bill Jenkins. There's a lot going on in computing. If you're not getting your computer science degree, kids, think about it again. He's Bill Jenkins. I'm Jeff Frick. You're watching theCUBE from Super Computing 2017. Thanks for watching. Thank you. (techno music)

>> Announcer: From Denver, Colorado, it's theCUBE. Covering Super Computing 2017 brought to you by Intel. (upbeat music) >> Hey, welcome back everybody. Jeffrey Frick here with theCube. We're in Denver, Colorado at Super Computing 17. I think it's the 20th year of the convention. 12,000 people. We've never been here before. It's pretty amazing. Amazing keynote, really talking about space, and really big, big, big computing projects, so, excited to be here, and we've got our first guest of the day. He's Bernard Friebe, he is the Senior Director of FPGA, I'll get that good by the end of the day, Software Solutions for Intel Programmable group. First off, welcome, Bernard. >> Thank you. I'm glad to be here. >> Absolutely. So, have you been to this conference before? >> Yeah, a couple of times before. It's always a big event. Always a big show for us, so I'm excited. >> Yeah, and it's different, too, cuz it's got a lot of academic influence, as well, as you walk around the outside. It's pretty hardcore. >> Yes, it's wonderful, and you see a lot of innovation going on, and we need to move fast. We need to move faster. That's what it is. And accelerate. >> And that's what you're all about, acceleration, so, Intel's making a lot of announcements, really, about acceleration at FPGA. For acceleration and in data centers and in big data, and all these big applications. So, explain just a little bit how that seed is evolving and what some of the recent announcements are all about. >> The world of computing must accelerate. I think we all agree on that. We all see that that's a key requirement. And FPGA's are a truly versatile, multi-function accelerator. It accelerates so many workloads in the high-performance computing space, may it be financial, genomics, oil and gas, data analytics, and the list goes on. Machine learning is a very big one. The list goes on and on. And, so, we're investing heavily in providing solutions which makes it much easier for our users to develop and deploy FPGA in a high-performance computing environment. >> You guys are taking a lot of steps to make the software programming at FPGA a lot easier, so you don't have to be a hardcore hardware engineer, so you can open it up to a broader ecosystem and get a broader solution set. Is that right? >> That's right, and it's not just the hardware. How do you unlock the benefits of FPGA as a versatile accelerator, so their parallelism, their ability to do real-time, low-latency, acceleration of many different workloads, and how do you enable that in an environment which is truly dynamic and multi-function, like a data center. And so, the product we've recently announced is the acceleration stack for xeon with FPGA, which enables that use more. >> So, what are the components for that stack? >> It starts with hardware. So, we are building a hardware accelerator card, it's a pc express plugin card, it's called programmable accelerator card. We have integrated solutions where you have everything on an FPGA in package, but what's common is a software framework solution stack, which sits on top of these different hardware implementation, which really makes it easy for a developer to develop an accelerator, for a user to then deploy that accelerator and run it in their environment, and it also enables a data center operator to basically enable the FPGA like any other computer resources by integrating it into their orchestration framework. So, multiple levels taking care of all those needs. >> It's interesting, because there's a lot of big trends that you guys are taking advantage of. Obviously, we're at Super Computing, but big data, streaming analytics, is all the rage now, so more data faster, reading it in real time, pumping it into the database in real time, and then, right around the corner, we have IoT and internet of things and all these connected devices. So the demand for increased speed, to get that data in, get that data processed, get the analytics back out, is only growing exponentially. >> That's right, and FPGAs, due to their flexibility, have distinct advantages there. The traditional model is look aside of offload, where you have a processor, and then you offload your tasks to your accelerator. The FPGA, with their flexible I/Os and flexible core can actually run directly in the data path, so that's what we call in-line processing. And what that allows people to do is, whatever the source is, may it be cameras, may it be storage, may it be through the network, through ethernet, can stream directly into the FPGA and do your acceleration as the data comes in in a streaming way. And FPGAs provide really unique advantages there versus other types of accelerators. Low-latency, very high band-width, and they're flexible in a sense that our customers can build different interfaces, different connectivity around those FPGAs. So, it's really amazing how versatile the usage of FPGA has become. >> It is pretty interesting, because you're using all the benefits that come from hardware, hardware-based solutions, which you just get a lot of benefits when things are hardwired, with the software component and enabling a broader ecosystem to write ready-made solutions and integrations to their existing solutions that they already have. Great approach. >> The acceleration stack provides a consistent interface to the developer and the user of the FPGA. What that allows our ecosystem and our customers to do is to define these accelerators based on this framework, and then they can easily migrate those between different hardware platforms, so we're building in future improvements of the solution, and the consistent interfaces then allow our customers and partners to build their software stacks on top of it. So, their investment, once they do it and we target our Arria 10 programmable accelerator card can easily be leveraged and moved forward into the next generation strategy, and beyond. We enable, really, and encourage a broad ecosystem, to build solutions. You'll see that here at the show, many partners now have demos, and they show their solutions built on Intel FPGA hardware and the acceleration stack. >> OK, so I'm going to put you on the spot. So, these are announced, what's the current state of the general availability? >> We're sampling now on the cards, the acceleration stack is available for delivery to customers. A lot of it is open source, by the way, so it can already be downloaded from GitHub And the partners are developing the solutions they are demonstrating today. The product will go into volume production in the first half of next year. So, we're very close. >> All right, very good. Well, Bernard, thanks for taking a few minutes to stop by. >> Oh, it's my pleasure. >> All right. He's Bernard, I'm Jeff. You're watching theCUBE from Super Computing 17. Thanks for watching. (upbeat music)

>> Welcome to the Intel AI Lounge. Today, we're very excited to share with you the Precision Medicine panel discussion. I'll be moderating the session. My name is Kay Erin. I'm the general manager of Health and Life Sciences at Intel. And I'm excited to share with you these three panelists that we have here. First is John Madison. He is a chief information medical officer and he is part of Kaiser Permanente. We're very excited to have you here. Thank you, John. >> Thank you. >> We also have Naveen Rao. He is the VP and general manager for the Artificial Intelligence Solutions at Intel. He's also the former CEO of Nervana, which was acquired by Intel. And we also have Bob Rogers, who's the chief data scientist at our AI solutions group. So, why don't we get started with our questions. I'm going to ask each of the panelists to talk, introduce themselves, as well as talk about how they got started with AI. So why don't we start with John? >> Sure, so can you hear me okay in the back? Can you hear? Okay, cool. So, I am a recovering evolutionary biologist and a recovering physician and a recovering geek. And I implemented the health record system for the first and largest region of Kaiser Permanente. And it's pretty obvious that most of the useful data in a health record, in lies in free text. So I started up a natural language processing team to be able to mine free text about a dozen years ago. So we can do things with that that you can't otherwise get out of health information. I'll give you an example. I read an article online from the New England Journal of Medicine about four years ago that said over half of all people who have had their spleen taken out were not properly vaccinated for a common form of pneumonia, and when your spleen's missing, you must have that vaccine or you die a very sudden death with sepsis. In fact, our medical director in Northern California's father died of that exact same scenario. So, when I read the article, I went to my structured data analytics team and to my natural language processing team and said please show me everybody who has had their spleen taken out and hasn't been appropriately vaccinated and we ran through about 20 million records in about three hours with the NLP team, and it took about three weeks with a structured data analytics team. That sounds counterintuitive but it actually happened that way. And it's not a competition for time only. It's a competition for quality and sensitivity and specificity. So we were able to indentify all of our members who had their spleen taken out, who should've had a pneumococcal vaccine. We vaccinated them and there are a number of people alive today who otherwise would've died absent that capability. So people don't really commonly associate natural language processing with machine learning, but in fact, natural language processing relies heavily and is the first really, highly successful example of machine learning. So we've done dozens of similar projects, mining free text data in millions of records very efficiently, very effectively. But it really helped advance the quality of care and reduce the cost of care. It's a natural step forward to go into the world of personalized medicine with the arrival of a 100-dollar genome, which is actually what it costs today to do a full genome sequence. Microbiomics, that is the ecosystem of bacteria that are in every organ of the body actually. And we know now that there is a profound influence of what's in our gut and how we metabolize drugs, what diseases we get. You can tell in a five year old, whether or not they were born by a vaginal delivery or a C-section delivery by virtue of the bacteria in the gut five years later. So if you look at the complexity of the data that exists in the genome, in the microbiome, in the health record with free text and you look at all the other sources of data like this streaming data from my wearable monitor that I'm part of a research study on Precision Medicine out of Stanford, there is a vast amount of disparate data, not to mention all the imaging, that really can collectively produce much more useful information to advance our understanding of science, and to advance our understanding of every individual. And then we can do the mash up of a much broader range of science in health care with a much deeper sense of data from an individual and to do that with structured questions and structured data is very yesterday. The only way we're going to be able to disambiguate those data and be able to operate on those data in concert and generate real useful answers from the broad array of data types and the massive quantity of data, is to let loose machine learning on all of those data substrates. So my team is moving down that pathway and we're very excited about the future prospects for doing that. >> Yeah, great. I think that's actually some of the things I'm very excited about in the future with some of the technologies we're developing. My background, I started actually being fascinated with computation in biological forms when I was nine. Reading and watching sci-fi, I was kind of a big dork which I pretty much still am. I haven't really changed a whole lot. Just basically seeing that machines really aren't all that different from biological entities, right? We are biological machines and kind of understanding how a computer works and how we engineer those things and trying to pull together concepts that learn from biology into that has always been a fascination of mine. As an undergrad, I was in the EE, CS world. Even then, I did some research projects around that. I worked in the industry for about 10 years designing chips, microprocessors, various kinds of ASICs, and then actually went back to school, quit my job, got a Ph.D. in neuroscience, computational neuroscience, to specifically understand what's the state of the art. What do we really understand about the brain? And are there concepts that we can take and bring back? Inspiration's always been we want to... We watch birds fly around. We want to figure out how to make something that flies. We extract those principles, and then build a plane. Don't necessarily want to build a bird. And so Nervana's really was the combination of all those experiences, bringing it together. Trying to push computation in a new a direction. Now, as part of Intel, we can really add a lot of fuel to that fire. I'm super excited to be part of Intel in that the technologies that we were developing can really proliferate and be applied to health care, can be applied to Internet, can be applied to every facet of our lives. And some of the examples that John mentioned are extremely exciting right now and these are things we can do today. And the generality of these solutions are just really going to hit every part of health care. I mean from a personal viewpoint, my whole family are MDs. I'm sort of the black sheep of the family. I don't have an MD. And it's always been kind of funny to me that knowledge is concentrated in a few individuals. Like you have a rare tumor or something like that, you need the guy who knows how to read this MRI. Why? Why is it like that? Can't we encapsulate that knowledge into a computer or into an algorithm, and democratize it. And the reason we couldn't do it is we just didn't know how. And now we're really getting to a point where we know how to do that. And so I want that capability to go to everybody. It'll bring the cost of healthcare down. It'll make all of us healthier. That affects everything about our society. So that's really what's exciting about it to me. >> That's great. So, as you heard, I'm Bob Rogers. I'm chief data scientist for analytics and artificial intelligence solutions at Intel. My mission is to put powerful analytics in the hands of every decision maker and when I think about Precision Medicine, decision makers are not just doctors and surgeons and nurses, but they're also case managers and care coordinators and probably most of all, patients. So the mission is really to put powerful analytics and AI capabilities in the hands of everyone in health care. It's a very complex world and we need tools to help us navigate it. So my background, I started with a Ph.D. in physics and I was computer modeling stuff, falling into super massive black holes. And there's a lot of applications for that in the real world. No, I'm kidding. (laughter) >> John: There will be, I'm sure. Yeah, one of these days. Soon as we have time travel. Okay so, I actually, about 1991, I was working on my post doctoral research, and I heard about neural networks, these things that could compute the way the brain computes. And so, I started doing some research on that. I wrote some papers and actually, it was an interesting story. The problem that we solved that got me really excited about neural networks, which have become deep learning, my office mate would come in. He was this young guy who was about to go off to grad school. He'd come in every morning. "I hate my project." Finally, after two weeks, what's your project? What's the problem? It turns out he had to circle these little fuzzy spots on these images from a telescope. So they were looking for the interesting things in a sky survey, and he had to circle them and write down their coordinates all summer. Anyone want to volunteer to do that? No? Yeah, he was very unhappy. So we took the first two weeks of data that he created doing his work by hand, and we trained an artificial neural network to do his summer project and finished it in about eight hours of computing. (crowd laughs) And so he was like yeah, this is amazing. I'm so happy. And we wrote a paper. I was the first author of course, because I was the senior guy at age 24. And he was second author. His first paper ever. He was very, very excited. So we have to fast forward about 20 years. His name popped up on the Internet. And so it caught my attention. He had just won the Nobel Prize in physics. (laughter) So that's where artificial intelligence will get you. (laughter) So thanks Naveen. Fast forwarding, I also developed some time series forecasting capabilities that allowed me to create a hedge fund that I ran for 12 years. After that, I got into health care, which really is the center of my passion. Applying health care to figuring out how to get all the data from all those siloed sources, put it into the cloud in a secure way, and analyze it so you can actually understand those cases that John was just talking about. How do you know that that person had had a splenectomy and that they needed to get that pneumovax? You need to be able to search all the data, so we used AI, natural language processing, machine learning, to do that and then two years ago, I was lucky enough to join Intel and, in the intervening time, people like Naveen actually thawed the AI winter and we're really in a spring of amazing opportunities with AI, not just in health care but everywhere, but of course, the health care applications are incredibly life saving and empowering so, excited to be here on this stage with you guys. >> I just want to cue off of your comment about the role of physics in AI and health care. So the field of microbiomics that I referred to earlier, bacteria in our gut. There's more bacteria in our gut than there are cells in our body. There's 100 times more DNA in that bacteria than there is in the human genome. And we're now discovering a couple hundred species of bacteria a year that have never been identified under a microscope just by their DNA. So it turns out the person who really catapulted the study and the science of microbiomics forward was an astrophysicist who did his Ph.D. in Steven Hawking's lab on the collision of black holes and then subsequently, put the other team in a virtual reality, and he developed the first super computing center and so how did he get an interest in microbiomics? He has the capacity to do high performance computing and the kind of advanced analytics that are required to look at a 100 times the volume of 3.2 billion base pairs of the human genome that are represented in the bacteria in our gut, and that has unleashed the whole science of microbiomics, which is going to really turn a lot of our assumptions of health and health care upside down. >> That's great, I mean, that's really transformational. So a lot of data. So I just wanted to let the audience know that we want to make this an interactive session, so I'll be asking for questions in a little bit, but I will start off with one question so that you can think about it. So I wanted to ask you, it looks like you've been thinking a lot about AI over the years. And I wanted to understand, even though AI's just really starting in health care, what are some of the new trends or the changes that you've seen in the last few years that'll impact how AI's being used going forward? >> So I'll start off. There was a paper published by a guy by the name of Tegmark at Harvard last summer that, for the first time, explained why neural networks are efficient beyond any mathematical model we predict. And the title of the paper's fun. It's called Deep Learning Versus Cheap Learning. So there were two sort of punchlines of the paper. One is is that the reason that mathematics doesn't explain the efficiency of neural networks is because there's a higher order of mathematics called physics. And the physics of the underlying data structures determined how efficient you could mine those data using machine learning tools. Much more so than any mathematical modeling. And so the second thing that was a reel from that paper is that the substrate of the data that you're operating on and the natural physics of those data have inherent levels of complexity that determine whether or not a 12th layer of neural net will get you where you want to go really fast, because when you do the modeling, for those math geeks in the audience, a factorial. So if there's 12 layers, there's 12 factorial permutations of different ways you could sequence the learning through those data. When you have 140 layers of a neural net, it's a much, much, much bigger number of permutations and so you end up being hardware-bound. And so, what Max Tegmark basically said is you can determine whether to do deep learning or cheap learning based upon the underlying physics of the data substrates you're operating on and have a good insight into how to optimize your hardware and software approach to that problem. >> So another way to put that is that neural networks represent the world in the way the world is sort of built. >> Exactly. >> It's kind of hierarchical. It's funny because, sort of in retrospect, like oh yeah, that kind of makes sense. But when you're thinking about it mathematically, we're like well, anything... The way a neural can represent any mathematical function, therfore, it's fully general. And that's the way we used to look at it, right? So now we're saying, well actually decomposing the world into different types of features that are layered upon each other is actually a much more efficient, compact representation of the world, right? I think this is actually, precisely the point of kind of what you're getting at. What's really exciting now is that what we were doing before was sort of building these bespoke solutions for different kinds of data. NLP, natural language processing. There's a whole field, 25 plus years of people devoted to figuring out features, figuring out what structures make sense in this particular context. Those didn't carry over at all to computer vision. Didn't carry over at all to time series analysis. Now, with neural networks, we've seen it at Nervana, and now part of Intel, solving customers' problems. We apply a very similar set of techniques across all these different types of data domains and solve them. All data in the real world seems to be hierarchical. You can decompose it into this hierarchy. And it works really well. Our brains are actually general structures. As a neuroscientist, you can look at different parts of your brain and there are differences. Something that takes in visual information, versus auditory information is slightly different but they're much more similar than they are different. So there is something invariant, something very common between all of these different modalities and we're starting to learn that. And this is extremely exciting to me trying to understand the biological machine that is a computer, right? We're figurig it out, right? >> One of the really fun things that Ray Chrisfall likes to talk about is, and it falls in the genre of biomimmicry, and how we actually replicate biologic evolution in our technical solutions so if you look at, and we're beginning to understand more and more how real neural nets work in our cerebral cortex. And it's sort of a pyramid structure so that the first pass of a broad base of analytics, it gets constrained to the next pass, gets constrained to the next pass, which is how information is processed in the brain. So we're discovering increasingly that what we've been evolving towards, in term of architectures of neural nets, is approximating the architecture of the human cortex and the more we understand the human cortex, the more insight we get to how to optimize neural nets, so when you think about it, with millions of years of evolution of how the cortex is structured, it shouldn't be a surprise that the optimization protocols, if you will, in our genetic code are profoundly efficient in how they operate. So there's a real role for looking at biologic evolutionary solutions, vis a vis technical solutions, and there's a friend of mine who worked with who worked with George Church at Harvard and actually published a book on biomimmicry and they wrote the book completely in DNA so if all of you have your home DNA decoder, you can actually read the book on your DNA reader, just kidding. >> There's actually a start up I just saw in the-- >> Read-Write DNA, yeah. >> Actually it's a... He writes something. What was it? (response from crowd member) Yeah, they're basically encoding information in DNA as a storage medium. (laughter) The company, right? >> Yeah, that same friend of mine who coauthored that biomimmicry book in DNA also did the estimate of the density of information storage. So a cubic centimeter of DNA can store an hexabyte of data. I mean that's mind blowing. >> Naveen: Highly done soon. >> Yeah that's amazing. Also you hit upon a really important point there, that one of the things that's changed is... Well, there are two major things that have changed in my perception from let's say five to 10 years ago, when we were using machine learning. You could use data to train models and make predictions to understand complex phenomena. But they had limited utility and the challenge was that if I'm trying to build on these things, I had to do a lot of work up front. It was called feature engineering. I had to do a lot of work to figure out what are the key attributes of that data? What are the 10 or 20 or 100 pieces of information that I should pull out of the data to feed to the model, and then the model can turn it into a predictive machine. And so, what's really exciting about the new generation of machine learning technology, and particularly deep learning, is that it can actually learn from example data those features without you having to do any preprogramming. That's why Naveen is saying you can take the same sort of overall approach and apply it to a bunch of different problems. Because you're not having to fine tune those features. So at the end of the day, the two things that have changed to really enable this evolution is access to more data, and I'd be curious to hear from you where you're seeing data come from, what are the strategies around that. So access to data, and I'm talking millions of examples. So 10,000 examples most times isn't going to cut it. But millions of examples will do it. And then, the other piece is the computing capability to actually take millions of examples and optimize this algorithm in a single lifetime. I mean, back in '91, when I started, we literally would have thousands of examples and it would take overnight to run the thing. So now in the world of millions, and you're putting together all of these combinations, the computing has changed a lot. I know you've made some revolutionary advances in that. But I'm curious about the data. Where are you seeing interesting sources of data for analytics? >> So I do some work in the genomics space and there are more viable permutations of the human genome than there are people who have ever walked the face of the earth. And the polygenic determination of a phenotypic expression translation, what are genome does to us in our physical experience in health and disease is determined by many, many genes and the interaction of many, many genes and how they are up and down regulated. And the complexity of disambiguating which 27 genes are affecting your diabetes and how are they up and down regulated by different interventions is going to be different than his. It's going to be different than his. And we already know that there's four or five distinct genetic subtypes of type II diabetes. So physicians still think there's one disease called type II diabetes. There's actually at least four or five genetic variants that have been identified. And so, when you start thinking about disambiguating, particularly when we don't know what 95 percent of DNA does still, what actually is the underlining cause, it will require this massive capability of developing these feature vectors, sometimes intuiting it, if you will, from the data itself. And other times, taking what's known knowledge to develop some of those feature vectors, and be able to really understand the interaction of the genome and the microbiome and the phenotypic data. So the complexity is high and because the variation complexity is high, you do need these massive members. Now I'm going to make a very personal pitch here. So forgive me, but if any of you have any role in policy at all, let me tell you what's happening right now. The Genomic Information Nondiscrimination Act, so called GINA, written by a friend of mine, passed a number of years ago, says that no one can be discriminated against for health insurance based upon their genomic information. That's cool. That should allow all of you to feel comfortable donating your DNA to science right? Wrong. You are 100% unprotected from discrimination for life insurance, long term care and disability. And it's being practiced legally today and there's legislation in the House, in mark up right now to completely undermine the existing GINA legislation and say that whenever there's another applicable statute like HIPAA, that the GINA is irrelevant, that none of the fines and penalties are applicable at all. So we need a ton of data to be able to operate on. We will not be getting a ton of data to operate on until we have the kind of protection we need to tell people, you can trust us. You can give us your data, you will not be subject to discrimination. And that is not the case today. And it's being further undermined. So I want to make a plea to any of you that have any policy influence to go after that because we need this data to help the understanding of human health and disease and we're not going to get it when people look behind the curtain and see that discrimination is occurring today based upon genetic information. >> Well, I don't like the idea of being discriminated against based on my DNA. Especially given how little we actually know. There's so much complexity in how these things unfold in our own bodies, that I think anything that's being done is probably childishly immature and oversimplifying. So it's pretty rough. >> I guess the translation here is that we're all unique. It's not just a Disney movie. (laughter) We really are. And I think one of the strengths that I'm seeing, kind of going back to the original point, of these new techniques is it's going across different data types. It will actually allow us to learn more about the uniqueness of the individual. It's not going to be just from one data source. They were collecting data from many different modalities. We're collecting behavioral data from wearables. We're collecting things from scans, from blood tests, from genome, from many different sources. The ability to integrate those into a unified picture, that's the important thing that we're getting toward now. That's what I think is going to be super exciting here. Think about it, right. I can tell you to visual a coin, right? You can visualize a coin. Not only do you visualize it. You also know what it feels like. You know how heavy it is. You have a mental model of that from many different perspectives. And if I take away one of those senses, you can still identify the coin, right? If I tell you to put your hand in your pocket, and pick out a coin, you probably can do that with 100% reliability. And that's because we have this generalized capability to build a model of something in the world. And that's what we need to do for individuals is actually take all these different data sources and come up with a model for an individual and you can actually then say what drug works best on this. What treatment works best on this? It's going to get better with time. It's not going to be perfect, because this is what a doctor does, right? A doctor who's very experienced, you're a practicing physician right? Back me up here. That's what you're doing. You basically have some categories. You're taking information from the patient when you talk with them, and you're building a mental model. And you apply what you know can work on that patient, right? >> I don't have clinic hours anymore, but I do take care of many friends and family. (laughter) >> You used to, you used to. >> I practiced for many years before I became a full-time geek. >> I thought you were a recovering geek. >> I am. (laughter) I do more policy now. >> He's off the wagon. >> I just want to take a moment and see if there's anyone from the audience who would like to ask, oh. Go ahead. >> We've got a mic here, hang on one second. >> I have tons and tons of questions. (crosstalk) Yes, so first of all, the microbiome and the genome are really complex. You already hit about that. Yet most of the studies we do are small scale and we have difficulty repeating them from study to study. How are we going to reconcile all that and what are some of the technical hurdles to get to the vision that you want? >> So primarily, it's been the cost of sequencing. Up until a year ago, it's $1000, true cost. Now it's $100, true cost. And so that barrier is going to enable fairly pervasive testing. It's not a real competitive market becaue there's one sequencer that is way ahead of everybody else. So the price is not $100 yet. The cost is below $100. So as soon as there's competition to drive the cost down, and hopefully, as soon as we all have the protection we need against discrimination, as I mentioned earlier, then we will have large enough sample sizes. And so, it is our expectation that we will be able to pool data from local sources. I chair the e-health work group at the Global Alliance for Genomics and Health which is working on this very issue. And rather than pooling all the data into a single, common repository, the strategy, and we're developing our five-year plan in a month in London, but the goal is to have a federation of essentially credentialed data enclaves. That's a formal method. HHS already does that so you can get credentialed to search all the data that Medicare has on people that's been deidentified according to HIPPA. So we want to provide the same kind of service with appropriate consent, at an international scale. And there's a lot of nations that are talking very much about data nationality so that you can't export data. So this approach of a federated model to get at data from all the countries is important. The other thing is a block-chain technology is going to be very profoundly useful in this context. So David Haussler of UC Santa Cruz is right now working on a protocol using an open block-chain, public ledger, where you can put out. So for any typical cancer, you may have a half dozen, what are called sematic variance. Cancer is a genetic disease so what has mutated to cause it to behave like a cancer? And if we look at those biologically active sematic variants, publish them on a block chain that's public, so there's not enough data there to reidentify the patient. But if I'm a physician treating a woman with breast cancer, rather than say what's the protocol for treating a 50-year-old woman with this cell type of cancer, I can say show me all the people in the world who have had this cancer at the age of 50, wit these exact six sematic variants. Find the 200 people worldwide with that. Ask them for consent through a secondary mechanism to donate everything about their medical record, pool that information of the core of 200 that exactly resembles the one sitting in front of me, and find out, of the 200 ways they were treated, what got the best results. And so, that's the kind of future where a distributed, federated architecture will allow us to query and obtain a very, very relevant cohort, so we can basically be treating patients like mine, sitting right in front of me. Same thing applies for establishing research cohorts. There's some very exciting stuff at the convergence of big data analytics, machine learning, and block chaining. >> And this is an area that I'm really excited about and I think we're excited about generally at Intel. They actually have something called the Collaborative Cancer Cloud, which is this kind of federated model. We have three different academic research centers. Each of them has a very sizable and valuable collection of genomic data with phenotypic annotations. So you know, pancreatic cancer, colon cancer, et cetera, and we've actually built a secure computing architecture that can allow a person who's given the right permissions by those organizations to ask a specific question of specific data without ever sharing the data. So the idea is my data's really important to me. It's valuable. I want us to be able to do a study that gets the number from the 20 pancreatic cancer patients in my cohort, up to the 80 that we have in the whole group. But I can't do that if I'm going to just spill my data all over the world. And there are HIPAA and compliance reasons for that. There are business reasons for that. So what we've built at Intel is this platform that allows you to do different kinds of queries on this genetic data. And reach out to these different sources without sharing it. And then, the work that I'm really involved in right now and that I'm extremely excited about... This also touches on something that both of you said is it's not sufficient to just get the genome sequences. You also have to have the phenotypic data. You have to know what cancer they've had. You have to know that they've been treated with this drug and they've survived for three months or that they had this side effect. That clinical data also needs to be put together. It's owned by other organizations, right? Other hospitals. So the broader generalization of the Collaborative Cancer Cloud is something we call the data exchange. And it's a misnomer in a sense that we're not actually exchanging data. We're doing analytics on aggregated data sets without sharing it. But it really opens up a world where we can have huge populations and big enough amounts of data to actually train these models and draw the thread in. Of course, that really then hits home for the techniques that Nervana is bringing to the table, and of course-- >> Stanford's one of your academic medical centers? >> Not for that Collaborative Cancer Cloud. >> The reason I mentioned Standford is because the reason I'm wearing this FitBit is because I'm a research subject at Mike Snyder's, the chair of genetics at Stanford, IPOP, intrapersonal omics profile. So I was fully sequenced five years ago and I get four full microbiomes. My gut, my mouth, my nose, my ears. Every three months and I've done that for four years now. And about a pint of blood. And so, to your question of the density of data, so a lot of the problem with applying these techniques to health care data is that it's basically a sparse matrix and there's a lot of discontinuities in what you can find and operate on. So what Mike is doing with the IPOP study is much the same as you described. Creating a highly dense longitudinal set of data that will help us mitigate the sparse matrix problem. (low volume response from audience member) Pardon me. >> What's that? (low volume response) (laughter) >> Right, okay. >> John: Lost the school sample. That's got to be a new one I've heard now. >> Okay, well, thank you so much. That was a great question. So I'm going to repeat this and ask if there's another question. You want to go ahead? >> Hi, thanks. So I'm a journalist and I report a lot on these neural networks, a system that's beter at reading mammograms than your human radiologists. Or a system that's better at predicting which patients in the ICU will get sepsis. These sort of fascinating academic studies that I don't really see being translated very quickly into actual hospitals or clinical practice. Seems like a lot of the problems are regulatory, or liability, or human factors, but how do you get past that and really make this stuff practical? >> I think there's a few things that we can do there and I think the proof points of the technology are really important to start with in this specific space. In other places, sometimes, you can start with other things. But here, there's a real confidence problem when it comes to health care, and for good reason. We have doctors trained for many, many years. School and then residencies and other kinds of training. Because we are really, really conservative with health care. So we need to make sure that technology's well beyond just the paper, right? These papers are proof points. They get people interested. They even fuel entire grant cycles sometimes. And that's what we need to happen. It's just an inherent problem, its' going to take a while. To get those things to a point where it's like well, I really do trust what this is saying. And I really think it's okay to now start integrating that into our standard of care. I think that's where you're seeing it. It's frustrating for all of us, believe me. I mean, like I said, I think personally one of the biggest things, I want to have an impact. Like when I go to my grave, is that we used machine learning to improve health care. We really do feel that way. But it's just not something we can do very quickly and as a business person, I don't actually look at those use cases right away because I know the cycle is just going to be longer. >> So to your point, the FDA, for about four years now, has understood that the process that has been given to them by their board of directors, otherwise known as Congress, is broken. And so they've been very actively seeking new models of regulation and what's really forcing their hand is regulation of devices and software because, in many cases, there are black box aspects of that and there's a black box aspect to machine learning. Historically, Intel and others are making inroads into providing some sort of traceability and transparency into what happens in that black box rather than say, overall we get better results but once in a while we kill somebody. Right? So there is progress being made on that front. And there's a concept that I like to use. Everyone knows Ray Kurzweil's book The Singularity Is Near? Well, I like to think that diadarity is near. And the diadarity is where you have human transparency into what goes on in the black box and so maybe Bob, you want to speak a little bit about... You mentioned that, in a prior discussion, that there's some work going on at Intel there. >> Yeah, absolutely. So we're working with a number of groups to really build tools that allow us... In fact Naveen probably can talk in even more detail than I can, but there are tools that allow us to actually interrogate machine learning and deep learning systems to understand, not only how they respond to a wide variety of situations but also where are there biases? I mean, one of the things that's shocking is that if you look at the clinical studies that our drug safety rules are based on, 50 year old white guys are the peak of that distribution, which I don't see any problem with that, but some of you out there might not like that if you're taking a drug. So yeah, we want to understand what are the biases in the data, right? And so, there's some new technologies. There's actually some very interesting data-generative technologies. And this is something I'm also curious what Naveen has to say about, that you can generate from small sets of observed data, much broader sets of varied data that help probe and fill in your training for some of these systems that are very data dependent. So that takes us to a place where we're going to start to see deep learning systems generating data to train other deep learning systems. And they start to sort of go back and forth and you start to have some very nice ways to, at least, expose the weakness of these underlying technologies. >> And that feeds back to your question about regulatory oversight of this. And there's the fascinating, but little known origin of why very few women are in clinical studies. Thalidomide causes birth defects. So rather than say pregnant women can't be enrolled in drug trials, they said any woman who is at risk of getting pregnant cannot be enrolled. So there was actually a scientific meritorious argument back in the day when they really didn't know what was going to happen post-thalidomide. So it turns out that the adverse, unintended consequence of that decision was we don't have data on women and we know in certain drugs, like Xanax, that the metabolism is so much slower, that the typical dosing of Xanax is women should be less than half of that for men. And a lot of women have had very serious adverse effects by virtue of the fact that they weren't studied. So the point I want to illustrate with that is that regulatory cycles... So people have known for a long time that was like a bad way of doing regulations. It should be changed. It's only recently getting changed in any meaningful way. So regulatory cycles and legislative cycles are incredibly slow. The rate of exponential growth in technology is exponential. And so there's impedance mismatch between the cycle time for regulation cycle time for innovation. And what we need to do... I'm working with the FDA. I've done four workshops with them on this very issue. Is that they recognize that they need to completely revitalize their process. They're very interested in doing it. They're not resisting it. People think, oh, they're bad, the FDA, they're resisting. Trust me, there's nobody on the planet who wants to revise these review processes more than the FDA itself. And so they're looking at models and what I recommended is global cloud sourcing and the FDA could shift from a regulatory role to one of doing two things, assuring the people who do their reviews are competent, and assuring that their conflicts of interest are managed, because if you don't have a conflict of interest in this very interconnected space, you probably don't know enough to be a reviewer. So there has to be a way to manage the conflict of interest and I think those are some of the keypoints that the FDA is wrestling with because there's type one and type two errors. If you underregulate, you end up with another thalidomide and people born without fingers. If you overregulate, you prevent life saving drugs from coming to market. So striking that balance across all these different technologies is extraordinarily difficult. If it were easy, the FDA would've done it four years ago. It's very complicated. >> Jumping on that question, so all three of you are in some ways entrepreneurs, right? Within your organization or started companies. And I think it would be good to talk a little bit about the business opportunity here, where there's a huge ecosystem in health care, different segments, biotech, pharma, insurance payers, etc. Where do you see is the ripe opportunity or industry, ready to really take this on and to make AI the competitive advantage. >> Well, the last question also included why aren't you using the result of the sepsis detection? We do. There were six or seven published ways of doing it. We did our own data, looked at it, we found a way that was superior to all the published methods and we apply that today, so we are actually using that technology to change clinical outcomes. As far as where the opportunities are... So it's interesting. Because if you look at what's going to be here in three years, we're not going to be using those big data analytics models for sepsis that we are deploying today, because we're just going to be getting a tiny aliquot of blood, looking for the DNA or RNA of any potential infection and we won't have to infer that there's a bacterial infection from all these other ancillary, secondary phenomenon. We'll see if the DNA's in the blood. So things are changing so fast that the opportunities that people need to look for are what are generalizable and sustainable kind of wins that are going to lead to a revenue cycle that are justified, a venture capital world investing. So there's a lot of interesting opportunities in the space. But I think some of the biggest opportunities relate to what Bob has talked about in bringing many different disparate data sources together and really looking for things that are not comprehensible in the human brain or in traditional analytic models. >> I think we also got to look a little bit beyond direct care. We're talking about policy and how we set up standards, these kinds of things. That's one area. That's going to drive innovation forward. I completely agree with that. Direct care is one piece. How do we scale out many of the knowledge kinds of things that are embedded into one person's head and get them out to the world, democratize that. Then there's also development. The underlying technology's of medicine, right? Pharmaceuticals. The traditional way that pharmaceuticals is developed is actually kind of funny, right? A lot of it was started just by chance. Penicillin, a very famous story right? It's not that different today unfortunately, right? It's conceptually very similar. Now we've got more science behind it. We talk about domains and interactions, these kinds of things but fundamentally, the problem is what we in computer science called NP hard, it's too difficult to model. You can't solve it analytically. And this is true for all these kinds of natural sorts of problems by the way. And so there's a whole field around this, molecular dynamics and modeling these sorts of things, that are actually being driven forward by these AI techniques. Because it turns out, our brain doesn't do magic. It actually doesn't solve these problems. It approximates them very well. And experience allows you to approximate them better and better. Actually, it goes a little bit to what you were saying before. It's like simulations and forming your own networks and training off each other. There are these emerging dynamics. You can simulate steps of physics. And you come up with a system that's much too complicated to ever solve. Three pool balls on a table is one such system. It seems pretty simple. You know how to model that, but it actual turns out you can't predict where a balls going to be once you inject some energy into that table. So something that simple is already too complex. So neural network techniques actually allow us to start making those tractable. These NP hard problems. And things like molecular dynamics and actually understanding how different medications and genetics will interact with each other is something we're seeing today. And so I think there's a huge opportunity there. We've actually worked with customers in this space. And I'm seeing it. Like Rosch is acquiring a few different companies in space. They really want to drive it forward, using big data to drive drug development. It's kind of counterintuitive. I never would've thought it had I not seen it myself. >> And there's a big related challenge. Because in personalized medicine, there's smaller and smaller cohorts of people who will benefit from a drug that still takes two billion dollars on average to develop. That is unsustainable. So there's an economic imperative of overcoming the cost and the cycle time for drug development. >> I want to take a go at this question a little bit differently, thinking about not so much where are the industry segments that can benefit from AI, but what are the kinds of applications that I think are most impactful. So if this is what a skilled surgeon needs to know at a particular time to care properly for a patient, this is where most, this area here, is where most surgeons are. They are close to the maximum knowledge and ability to assimilate as they can be. So it's possible to build complex AI that can pick up on that one little thing and move them up to here. But it's not a gigantic accelerator, amplifier of their capability. But think about other actors in health care. I mentioned a couple of them earlier. Who do you think the least trained actor in health care is? >> John: Patients. >> Yes, the patients. The patients are really very poorly trained, including me. I'm abysmal at figuring out who to call and where to go. >> Naveen: You know as much the doctor right? (laughing) >> Yeah, that's right. >> My doctor friends always hate that. Know your diagnosis, right? >> Yeah, Dr. Google knows. So the opportunities that I see that are really, really exciting are when you take an AI agent, like sometimes I like to call it contextually intelligent agent, or a CIA, and apply it to a problem where a patient has a complex future ahead of them that they need help navigating. And you use the AI to help them work through. Post operative. You've got PT. You've got drugs. You've got to be looking for side effects. An agent can actually help you navigate. It's like your own personal GPS for health care. So it's giving you the inforamation that you need about you for your care. That's my definition of Precision Medicine. And it can include genomics, of course. But it's much bigger. It's that broader picture and I think that a sort of agent way of thinking about things and filling in the gaps where there's less training and more opportunity, is very exciting. >> Great start up idea right there by the way. >> Oh yes, right. We'll meet you all out back for the next start up. >> I had a conversation with the head of the American Association of Medical Specialties just a couple of days ago. And what she was saying, and I'm aware of this phenomenon, but all of the medical specialists are saying, you're killing us with these stupid board recertification trivia tests that you're giving us. So if you're a cardiologist, you have to remember something that happens in one in 10 million people, right? And they're saying that irrelevant anymore, because we've got advanced decision support coming. We have these kinds of analytics coming. Precisely what you're saying. So it's human augmentation of decision support that is coming at blazing speed towards health care. So in that context, it's much more important that you have a basic foundation, you know how to think, you know how to learn, and you know where to look. So we're going to be human-augmented learning systems much more so than in the past. And so the whole recertification process is being revised right now. (inaudible audience member speaking) Speak up, yeah. (person speaking) >> What makes it fathomable is that you can-- (audience member interjects inaudibly) >> Sure. She was saying that our brain is really complex and large and even our brains don't know how our brains work, so... are there ways to-- >> What hope do we have kind of thing? (laughter) >> It's a metaphysical question. >> It circles all the way down, exactly. It's a great quote. I mean basically, you can decompose every system. Every complicated system can be decomposed into simpler, emergent properties. You lose something perhaps with each of those, but you get enough to actually understand most of the behavior. And that's really how we understand the world. And that's what we've learned in the last few years what neural network techniques can allow us to do. And that's why our brain can understand our brain. (laughing) >> Yeah, I'd recommend reading Chris Farley's last book because he addresses that issue in there very elegantly. >> Yeah we're seeing some really interesting technologies emerging right now where neural network systems are actually connecting other neural network systems in networks. You can see some very compelling behavior because one of the things I like to distinguish AI versus traditional analytics is we used to have question-answering systems. I used to query a database and create a report to find out how many widgets I sold. Then I started using regression or machine learning to classify complex situations from this is one of these and that's one of those. And then as we've moved more recently, we've got these AI-like capabilities like being able to recognize that there's a kitty in the photograph. But if you think about it, if I were to show you a photograph that happened to have a cat in it, and I said, what's the answer, you'd look at me like, what are you talking about? I have to know the question. So where we're cresting with these connected sets of neural systems, and with AI in general, is that the systems are starting to be able to, from the context, understand what the question is. Why would I be asking about this picture? I'm a marketing guy, and I'm curious about what Legos are in the thing or what kind of cat it is. So it's being able to ask a question, and then take these question-answering systems, and actually apply them so that's this ability to understand context and ask questions that we're starting to see emerge from these more complex hierarchical neural systems. >> There's a person dying to ask a question. >> Sorry. You have hit on several different topics that all coalesce together. You mentioned personalized models. You mentioned AI agents that could help you as you're going through a transitionary period. You mentioned data sources, especially across long time periods. Who today has access to enough data to make meaningful progress on that, not just when you're dealing with an issue, but day-to-day improvement of your life and your health? >> Go ahead, great question. >> That was a great question. And I don't think we have a good answer to it. (laughter) I'm sure John does. Well, I think every large healthcare organization and various healthcare consortiums are working very hard to achieve that goal. The problem remains in creating semantic interoperatability. So I spent a lot of my career working on semantic interoperatability. And the problem is that if you don't have well-defined, or self-defined data, and if you don't have well-defined and documented metadata, and you start operating on it, it's real easy to reach false conclusions and I can give you a classic example. It's well known, with hundreds of studies looking at when you give an antibiotic before surgery and how effective it is in preventing a post-op infection. Simple question, right? So most of the literature done prosectively was done in institutions where they had small sample sizes. So if you pool that, you get a little bit more noise, but you get a more confirming answer. What was done at a very large, not my own, but a very large institution... I won't name them for obvious reasons, but they pooled lots of data from lots of different hospitals, where the data definitions and the metadata were different. Two examples. When did they indicate the antibiotic was given? Was it when it was ordered, dispensed from the pharmacy, delivered to the floor, brought to the bedside, put in the IV, or the IV starts flowing? Different hospitals used a different metric of when it started. When did surgery occur? When they were wheeled into the OR, when they were prepped and drapped, when the first incision occurred? All different. And they concluded quite dramatically that it didn't matter when you gave the pre-op antibiotic and whether or not you get a post-op infection. And everybody who was intimate with the prior studies just completely ignored and discounted that study. It was wrong. And it was wrong because of the lack of commonality and the normalization of data definitions and metadata definitions. So because of that, this problem is much more challenging than you would think. If it were so easy as to put all these data together and operate on it, normalize and operate on it, we would've done that a long time ago. It's... Semantic interoperatability remains a big problem and we have a lot of heavy lifting ahead of us. I'm working with the Global Alliance, for example, of Genomics and Health. There's like 30 different major ontologies for how you represent genetic information. And different institutions are using different ones in different ways in different versions over different periods of time. That's a mess. >> Our all those issues applicable when you're talking about a personalized data set versus a population? >> Well, so N of 1 studies and single-subject research is an emerging field of statistics. So there's some really interesting new models like step wedge analytics for doing that on small sample sizes, recruiting people asynchronously. There's single-subject research statistics. You compare yourself with yourself at a different point in time, in a different context. So there are emerging statistics to do that and as long as you use the same sensor, you won't have a problem. But people are changing their remote sensors and you're getting different data. It's measured in different ways with different sensors at different normalization and different calibration. So yes. It even persists in the N of 1 environment. >> Yeah, you have to get started with a large N that you can apply to the N of 1. I'm actually going to attack your question from a different perspective. So who has the data? The millions of examples to train a deep learning system from scratch. It's a very limited set right now. Technology such as the Collaborative Cancer Cloud and The Data Exchange are definitely impacting that and creating larger and larger sets of critical mass. And again, not withstanding the very challenging semantic interoperability questions. But there's another opportunity Kay asked about what's changed recently. One of the things that's changed in deep learning is that we now have modules that have been trained on massive data sets that are actually very smart as certain kinds of problems. So, for instance, you can go online and find deep learning systems that actually can recognize, better than humans, whether there's a cat, dog, motorcycle, house, in a photograph. >> From Intel, open source. >> Yes, from Intel, open source. So here's what happens next. Because most of that deep learning system is very expressive. That combinatorial mixture of features that Naveen was talking about, when you have all these layers, there's a lot of features there. They're actually very general to images, not just finding cats, dogs, trees. So what happens is you can do something called transfer learning, where you take a small or modest data set and actually reoptimize it for your specific problem very, very quickly. And so we're starting to see a place where you can... On one end of the spectrum, we're getting access to the computing capabilities and the data to build these incredibly expressive deep learning systems. And over here on the right, we're able to start using those deep learning systems to solve custom versions of problems. Just last weekend or two weekends ago, in 20 minutes, I was able to take one of those general systems and create one that could recognize all different kinds of flowers. Very subtle distinctions, that I would never be able to know on my own. But I happen to be able to get the data set and literally, it took 20 minutes and I have this vision system that I could now use for a specific problem. I think that's incredibly profound and I think we're going to see this spectrum of wherever you are in your ability to get data and to define problems and to put hardware in place to see really neat customizations and a proliferation of applications of this kind of technology. >> So one other trend I think, I'm very hopeful about it... So this is a hard problem clearly, right? I mean, getting data together, formatting it from many different sources, it's one of these things that's probably never going to happen perfectly. But one trend I think that is extremely hopeful to me is the fact that the cost of gathering data has precipitously dropped. Building that thing is almost free these days. I can write software and put it on 100 million cell phones in an instance. You couldn't do that five years ago even right? And so, the amount of information we can gain from a cell phone today has gone up. We have more sensors. We're bringing online more sensors. People have Apple Watches and they're sending blood data back to the phone, so once we can actually start gathering more data and do it cheaper and cheaper, it actually doesn't matter where the data is. I can write my own app. I can gather that data and I can start driving the correct inferences or useful inferences back to you. So that is a positive trend I think here and personally, I think that's how we're going to solve it, is by gathering from that many different sources cheaply. >> Hi, my name is Pete. I've very much enjoyed the conversation so far but I was hoping perhaps to bring a little bit more focus into Precision Medicine and ask two questions. Number one, how have you applied the AI technologies as you're emerging so rapidly to your natural language processing? I'm particularly interested in, if you look at things like Amazon Echo or Siri, or the other voice recognition systems that are based on AI, they've just become incredibly accurate and I'm interested in specifics about how I might use technology like that in medicine. So where would I find a medical nomenclature and perhaps some reference to a back end that works that way? And the second thing is, what specifically is Intel doing, or making available? You mentioned some open source stuff on cats and dogs and stuff but I'm the doc, so I'm looking at the medical side of that. What are you guys providing that would allow us who are kind of geeks on the software side, as well as being docs, to experiment a little bit more thoroughly with AI technology? Google has a free AI toolkit. Several other people have come out with free AI toolkits in order to accelerate that. There's special hardware now with graphics, and different processors, hitting amazing speeds. And so I was wondering, where do I go in Intel to find some of those tools and perhaps learn a bit about the fantastic work that you guys are already doing at Kaiser? >> Let me take that first part and then we'll be able to talk about the MD part. So in terms of technology, this is what's extremely exciting now about what Intel is focusing on. We're providing those pieces. So you can actually assemble and build the application. How you build that application specific for MDs and the use cases is up to you or the one who's filling out the application. But we're going to power that technology for multiple perspectives. So Intel is already the main force behind The Data Center, right? Cloud computing, all this is already Intel. We're making that extremely amenable to AI and setting the standard for AI in the future, so we can do that from a number of different mechanisms. For somebody who wants to develop an application quickly, we have hosted solutions. Intel Nervana is kind of the brand for these kinds of things. Hosted solutions will get you going very quickly. Once you get to a certain level of scale, where costs start making more sense, things can be bought on premise. We're supplying that. We're also supplying software that makes that transition essentially free. Then taking those solutions that you develop in the cloud, or develop in The Data Center, and actually deploying them on device. You want to write something on your smartphone or PC or whatever. We're actually providing those hooks as well, so we want to make it very easy for developers to take these pieces and actually build solutions out of them quickly so you probably don't even care what hardware it's running on. You're like here's my data set, this is what I want to do. Train it, make it work. Go fast. Make my developers efficient. That's all you care about, right? And that's what we're doing. We're taking it from that point at how do we best do that? We're going to provide those technologies. In the next couple of years, there's going to be a lot of new stuff coming from Intel. >> Do you want to talk about AI Academy as well? >> Yeah, that's a great segway there. In addition to this, we have an entire set of tutorials and other online resources and things we're going to be bringing into the academic world for people to get going quickly. So that's not just enabling them on our tools, but also just general concepts. What is a neural network? How does it work? How does it train? All of these things are available now and we've made a nice, digestible class format that you can actually go and play with. >> Let me give a couple of quick answers in addition to the great answers already. So you're asking why can't we use medical terminology and do what Alexa does? Well, no, you may not be aware of this, but Andrew Ian, who was the AI guy at Google, who was recruited by Google, they have a medical chat bot in China today. I don't speak Chinese. I haven't been able to use it yet. There are two similar initiatives in this country that I know of. There's probably a dozen more in stealth mode. But Lumiata and Health Cap are doing chat bots for health care today, using medical terminology. You have the compound problem of semantic normalization within language, compounded by a cross language. I've done a lot of work with an international organization called Snowmed, which translates medical terminology. So you're aware of that. We can talk offline if you want, because I'm pretty deep into the semantic space. >> Go google Intel Nervana and you'll see all the websites there. It's intel.com/ai or nervanasys.com. >> Okay, great. Well this has been fantastic. I want to, first of all, thank all the people here for coming and asking great questions. I also want to thank our fantastic panelists today. (applause) >> Thanks, everyone. >> Thank you. >> And lastly, I just want to share one bit of information. We will have more discussions on AI next Tuesday at 9:30 AM. Diane Bryant, who is our general manager of Data Centers Group will be here to do a keynote. So I hope you all get to join that. Thanks for coming. (applause) (light electronic music)