 Live from Las Vegas, it's theCUBE covering Dell EMC World 2017. Brought to you by Dell EMC. And welcome back here on theCUBE. We continue our coverage of Dell EMC World 2017. We're live here at the Sands Expo, the Venetian beautiful facility, great show, 13,000 plus attendees, a record number there, turning out for this three-day event. Along with Paul, Gil and I'm John Walls. Nice to have you with us. And we're joined now by, I think a gentleman that we can all relate to, at least from the MasterCard perspective, the Credit Card perspective, Nick Kukuru, the Vice President of Big Data Practice, Nick. Thanks for being with us, we appreciate the time. It's a pleasure to be here, especially at this particular event. Yeah, good. And Tony Parkinson, who is the VP of Sales and Systems Engineering at Dell EMC. And Tony, good to see you, sir. Quite to be here, John, thank you. So let's relate first off before we get into what machine learning and what you guys are doing with it. Let's just set the big picture here, credit cards, right? Paul and I were talking about it. We've been victimized, unfortunately. Many people are. How big is the fraud business, if you will, the detection business for you? How critical is that for you and MasterCard? Well, for us at MasterCard, the biggest thing is always making sure that people that use a MasterCard are protected. So we at MasterCard take that particular, whether it's fraud, anti-money wandering, that we take that individual and we are stewards of what that person has entrusted us with. We're stewards of their data. We're stewards of how their experience that they have with it. So we want them always to make sure from a fraud perspective, especially that they're protected. We're doing everything you can to make sure the person who uses that card is the person that has that card, right? So again, when you look at fraud, it's one of the biggest plays that the cyber attacks happen because it's an easy way to make money quickly, right? And for us, it's a ability to say how can we stay one step ahead because the fraudsters themselves are just as smart as we are. They're graduating from some of the top universities worldwide. They are carrying and going through a lot. They're doing big data and analytics themselves. So they are getting more sophisticated. They are getting more complicated. They are absolutely, this is a big business for fraudsters worldwide. So in this kind of cyber arms race that you're talking about here a bit, machine learning, how does that come into play with what you're doing and why does that give you, you think, maybe a leg up? Well, when you take a look at machine learning and what its potential is, right? So for us, machine learning allows us to actually bring in the data so much faster. Again, our goal is to know that when you're making a transaction, we don't want to impede that transaction you're going to have with the merchant. We want that experience to go smoothly, but we also want to make sure that transaction is a valid one. So when you look at machine learning and its ability to start to actually do the speed in milliseconds to understand, to apply rules in regards to this is what a fraudulent transaction looks like, this is what it doesn't look like. The ability to get closer to the transaction with those rule sets or the set of rules is really a big deal because again, it's milliseconds now. It's not after the fact. It's not chasing, it's now I am going to prevent this up front if I can. And also with machine learning, the goal is for us we call them false positives. It allows us to reduce false positives again to make sure that within our data and when someone's doing a transaction that card goes through. We don't want it to be like you've shown up in Chicago, now you're in London and your card shuts down. We want to be able to say that's a valid transaction. So again, the way we can alert people is this really you, the way we can interact with them to make sure those transactions are all valid. Have you been able to introduce new variables into the machine learning process because you have this capability now? Yeah, the ability when you look at machine learning there's two ways to look at it. There's supervised learning, there's unsupervised learning. And when we looked at it as we actually do we're kind of a hybrid, which is semi-supervised. So we want to take what we know about fraud. Again, master card, we know a lot. I mean, we've got close to a million rules somewhere that we can apply to transactions in different parts of the world. But the goal is to take that supervised learning as it teaches the machines what it needs to know, but then the unsupervised, what are those new points of fraud coming through? That really helps us move that needle. So again, for us when we take a look at machine learning it's that ability to say we want the machine to learn, we also want to supervise it to an extent. So again, we do a kind of, as we experiment with machine learning we want to make sure that we're always kind of looking at what we're getting in results to make sure it's proper. And that the results are positive. Tony, I have to admit I would not have put big data analytics and Dell together in the same sentence previously, what are you doing in this area that we need to know about? No, absolutely. I'm saying if you look at the platform infrastructure we're providing for Nick to achieve those millisecond response times that's where the technologies that Dell has been our latest generation of compute platforms, our 14G servers, which announced here this week, as well as the Iceland technologies from a storage perspective. We're really working very close partnership with partners like Hardera and Vidya to integrate that platform and make his ability to go chase that experience down so the machine can learn faster. That's really our engagement with great customers like last week. What platforms are you using for the machine learning? Well, when you take a look at it, again, it's not necessarily the platform itself obviously we have done and we use it. There's other ones that we use but the thing about it is pretty much Hadoop has helped us move the needle. And that's because Hadoop's first wave was all about speed and flexibility. The next wave that we're starting to see is about machine learning in itself. The ability to learn what you know so you can apply it to larger data sets. So where Dell comes into play is to provide us the ability to have commodities, quote unquote, please don't hurt me, but commodity type hardware to be able to process so you lower cost of ownership which gives us the ability to process faster, larger sets of data. And then also when we take a look at that when we look at partners like at Cloudera and you look at partners such as even the other distribution systems out there whether it's work and works or map are the ability to sit back and say they're creating work benches but now go across and take these silo tools that we're in Hadoop and now are creating a way to actually move them together seamlessly to again take the power of each individual tool and then magnify it. And again when you got Python running you've got Spark and you've got Solar all working together. Again that speed, that efficiency is just opens up a lot more opportunities. Now we're seeing these vendors and MapR just introduced an edge-oriented version of their suite. We're seeing them begin to think about how does big data extend out to the edge? And we were talking earlier about the delay and what an issue of the delay is in customer experience and improving credit card transactions. How do you see these architectures evolving with will more intelligence move out to the edge? Yeah and that's actually what you want to get to that experience because when you move it to the edge you're getting closer to the consumer, the customer itself, right? And that's what you want to be able to do is now they can have an interaction, a real relationship with you. And when you move it to the edge and you get closer and closer, the opportunity now allows you to help that customer get what they're looking for, right? Again, the question is always when you do that is how you're interacting with that customer. Is it for their benefit or is it for your benefit? So I think as you look at from a cultural standpoint as you look at and you get to the edge the question is how does the customer benefit more than potentially the vendor themselves, right? Because you don't always be putting out offers. What you want that customer is probably to get knowledge because the customer experience is about an emotional attachment and an intellectual attachment. And you want that person to be both stimulated and have that experience with you versus all you're trying to do is sell me another widget. So Tony, if you're looking to offer some advice to somebody about embarking on the machine learning journey what would that be? Yeah, I think certainly it is a journey. And I think the deployment of a machine learning is part of that journey. So start off with some data capture. Customers have been doing data capture years. The type of data is changing. So the volume of data, I've got my traditional data now I've got my unstructured data. But then moving into the world of machine learning you're seeing new tools. And I think there's been a fear that I've got to have a bunch of PhDs to help me on that journey. There's tools coming now open source tools from Cloud Era that will allow you to start to understand what is my data sets. And I think we talked about this last night is start small. Don't try to solve my entire data problem. Start off with a small chunk that I can then measure results and then start building my skill set and my knowledge and start adding new parameters versus we're going to have this massive project and we're not really ready for it. I think that's the advice we would see is to start small and grow. And I think that's been Nick's experience. Well, one of the things is when you do start small and you get focused, you also get the organization of buying in. One of the things that machine learning if there's an agreement is a cultural acceptance. It is a black box. Literally it's a black server box, right? That sits there. So now it's like, how do you demystify this? How do you prove to the organization what they're seeing will help solve their problems? That's when you have to tie it to a metric. That's where you start to be able to say let's get very focused on what we want to try that this machine predicts for us, right? How this machine learns for us, whether it's fraud or new sales or how to get more efficient on a production line. If you start small at that point and you were able to measure it and you're able to actually sit back and say the organization starts to get more and more trusting in what they're seeing. Because again, the impeachment is it's a black box. But we're typically seeing the start is in sales and marketing. Those are the, they're really good logical starting points around that journey. But it is a cultural buyer though, right? Absolutely. You've really got to be all in. Yep. And that's why if you're going to start, I usually, one of the suggestions we'll work with our clients more in MasterCard, MasterCard advisors, is you probably want to go supervised. Supervised machine learning, meaning you know what the results are going to be or should be. And then the system's like, oh, that's what it should be. I see it's happening, it's right. Organization starts to buy in and then you can start to move to that semi-supervised. Let some of that unsupervised learning happen, be able to alert you, create the anomaly detections and then it gives you the privilege to go forward. But really it's starting as like, if you got to get the buy in, supervised machine learning is the way to go, to get it and then eventually work your way through. So the words that place them are, walk before you run. Absolutely. Take it slow. Gentlemen, thanks for being with us here. Thank you. Appreciate it. Good to see you both. Cheers. We will continue live here from Las Vegas, Dell EMC World 2017, right after this.