The following is an unauthorised transcript of the Podcast between Rob Telson Brainchip and Rob Lincourt Dell Technologies on 3.5.21. It is intended to act as anaide-mémoire to the actual podcast which is available on the Brainchip web site under LEARN - Podcasts. Regards FF. AKIDA Ballista
This is the Brainchip podcast. Hear from ourthought leaders about neuromorphic computing, beneficial Ai and how Brainchip’sAKIDA is pushing Ai to the Edge. This podcast is for investors, practitionersand anyone interested in the future of Ai – 3 MAY, 2021
Hi all I’m Rob Telson Vice President of Sales and Marketing atBrainchip. Welcome and thank you for joining our sixth episode of ourBrainchip podcast series. These events are structured to provide currentand future investors and those interested in Brainchip technology as a path tobetter understand who we are, what we are doing, and where we are going.
Just to recap our first four episodes focused onBrainchip our company history, who we are, where we are going and theperspectives from our founders, the BRAIN, Peter van der Made, and the CHIP,Anil Mankar.
Our fifth episode we journeyed into the ecosystemand focused on the outside looking in, in a discussion with Alex Divinsky alsoknown as ‘Ticker Symbol U the channel that invests in you’ based on YouTube.
If you have not listened to any of these podcastsplease go to our website at SHAPE \* MERGEFORMAT www.brainchip.com clickon the Learn tab and select podcasts. You can also listen to any of thesepodcasts on your favourite PC Platform.
_________________________________________________________________________________
Rob Telson: Today we will continue with our outsidelooking in theme by having a discussion with Rob Lincourt distinguishedengineer at Dell Technologies. Rob has spent over 20 years with Dellfocused on new technology introduction.
Rob on behalf of Brainchip and our listeners thankyou very much for being with us today.
Rob Lincourt: Thank you for the invitation and having me. It’s apleasure.
Rob Telson: Well, you know I can only imagine the types oftechnology and types of innovations you have experienced in your career. You are a true technologist who has been involved in over a dozen patents andtruly epitomises the phrase “out of the box thinker.’ I had theopportunity to read your paper titled ‘Security reshaped in the DigitalTransformation Era’. It provided great insight on how we can protectourselves going through these rapidly evolving times. Rob why don’t you take amoment and provide our listeners with a bit of background on yourself?
Rob Lincourt: I work in the Global CTO Office and what wedo there is all about looking at emerging technologies, understanding them,understanding the impact to the industry, as well as to Dell technology.
So, I am here today with you from a Dell Technologypoint of view not a specific Dell product line view and we won’t talk anythingabout the product lines or any of the other Dell Technologies companies so itwill just strictly be a view of the technology, of where the industry is andthe impact that we have. Again, the roll here at Dell is about understandinghow new technology is going to affect us, the people, and where we are going to,going forward from, from here is what’s new.
There are a bunch of strategic areas that Dell islooking at and we will try to tie some of those into, to what we are talkingabout today.
Rob Telson: That’s great and I think it’s exciting that youhave that opportunity to focus on some of those strategic areas. As you look atthe technology and how everything is evolving what aspects currently have yourinterest and why?
Rob Lincourt: The list is long for this short conversation werehaving and I guess its really about the intersection of a lot of theseverticals and horizontals that people have, are talking in the world betweenedge computing and Ai.
Ai is really the driver of why we are doing Edgecomputing. It is so important to act on the data and get the decisions, getactions out of it as soon as you can. Especially when you start thinkingabout autonomous vehicles. When your start thinking about how hospitals of thefuture will work and how we are going to interact with it. I was just at aconference, MTac, the technology Review from MIT and we spent a wholeafternoon just talking about how people are going to consume Ai, and so how doI have to interact with the Ai to get you Rob, to believe me, and to understandthe result, and trust in the results that I am giving you.
So, there is this whole new human dynamic cominginto the Ai space so all of those spaces together are going to need obviouslynone of this can happen without cloud in some way shape or form but we feelthat data can’t move all the way back to the cloud every single time for everysingle decision. The latencies just aren’t there. Now the cloudmight slide out a little bit, right, there is going to be this huge concept ofthe Edge and these notions that Edge is everywhere. And the funny thingis a lot of the topics we talk about is all about perspective.
Where is the Edge from where I am standing? Is itat the device and the device can be a sensor, camera, monitor, a microphone.Then it can go all the way back to a car. So, imagine a car being a device andnow moving all the way back to the Telco network, you know I can do processingthere for now, especially with 5G and the ability to move data and segment dateand all this other fun stuff. Smart cities start to become a reality,smart buildings all these things acting together are all Edges, so depends onwhere you are standing what you see.
Rob Telson: This is a really good point, and this is one ofthose areas where Brainchip would differentiate ourselves in our opinion andwhere the market is today and where Ai is moving very fast as you havehighlighted then the fact Edge becomes a really big part of thediscussion. And a lot of companies with their deep learning acceleratorslook at the Edge, the Edge of the cloud and the Edge of the data centre and withBrainchip and our AKIDA product we are focused on the Edge at the sensor, orthe device and there would be billions and billions of IOT devices out thereright now and its only increasing at a rapid rate each year and we look at theEdge and focus on that in a different fashion than what most companies nowapproach the Edge as it is today. How do you see neuromorphic processingchanging the face of Ai?
Rob Lincourt: So, I think it’s going to change the way we look,like you said, at the big neural nets and the amount of training involved andthe data that’s needed. If I can start to minimise that traffic or learn alittle bit at the Edge, transfer some of that back or share it throughout anecosystem it’s going to help tremendously. The other big part of it is,the one thing you didn’t mention, is the power constraint at all thosedifferent points that we talked about you know we have to now start concerningourselves what can we consume at different steps at that Edge environmentand I think everything I have been seeing about neuromorphic is it is very lowpowered, because I can get a couple of cores, or I can get a chip with a couplemore cores, and I can start to stitch them together. That to me is theinteresting thing and I also get the ability to reconfigure it so I use it forone set of things and then I can bring and almost I don’t want to say instantly,that’s a precise word, but very quickly I can switch it to now and startrunning a different neural net. You know I get all those benefits out of itthat, and that’s when I look to start seeing the benefits of neuromorphic, isthat power in having a very nice linear power as we add more cores.
Rob Telson: Great points and power is definitely an area wefeel very strongly that we can make a massive impact and the results that wereseeing with our current chip and some of the analysis that we have done so farit is working out that way that there is going to be this massive powerreduction which is going to end up giving companies a lot of flexibility howthey want to move forward with Ai and what they want to do.
What is it about Brainchipthat interests you?
Rob Lincourt: So again, it was that power, so you published somepower numbers very early on and those were very encouraging and intriguing tous and again Dell Technologies has a plethora of products that are throughoutthat Edge ecosystem, and yea to me that having Brainchip, as potentially aspart of that environment, one it gives us a couple of advantages.
One, power again, right, I don’t want to drainbatteries, I want to operate more on batteries, you know the other this is ofthe other environmental aspects. I don’t want to use a lot of power. I don’twant to create a lot of heat. I will be able to do all these otherenvironmentals, I have to interact with or worry about.
The other thing is again that reconfigurabilityand, right, that ability to move it back and forth and again at DellTechnologies we really want to understand where things are going and wherethese different technologies can be useful and where they can’t be. You also have to look at that as well you knowif you want to look at what you can do with neuromorphic just look at what theM5 has from Apple. Look at what Snapdragon has for Qualcomm. They all have these little neurosections whatever they want to call it but their starting to learn that theycan now off load all sorts of new functionality to it that they didn’t reallythink it would matter before so to me Brainchip either being a companion chipto a camera where I can do some of that object detection, face detection fordoorbells, automatic doors, for whatever you want. Then it allows me to go into a whole new spacewhere I have this heterogeneous computer where it’s on the same die or spreadthroughout multiple systems. It just depends on the scale that you can operateat and again that’s what we ‘kinda’ feel and that’s ‘kinda’ what we have beentrying to play around with, what is that scale that we can achieve.
Rob Telson: That’s an area where we see a lot of opportunitiesand it expands from as you said helping out being an engine that helps out in alarger processing area, as a companion as you called it, all the way throughand being the lead engine and processing the data on the Edge as we see Aievolve and we see companies such as Dell start to make progress and move downthese paths we will just have to see how a lot of these Ai technologies evolvewith it, right?
Rob Lincourt: Yep.
Rob Telson: So here at Brainchip we talk a lot about ways thatour technology can be beneficial for society at large. We call it beneficialAi, and through our approach of focusing on the five sensing modalities. Don’tknow if you have seen our robot slide its vision, hearing, touch, smell andtaste. What areas do you feel Ai can have the most beneficial impact onsociety?
Rob Lincourt: Interesting question that can go a lot of differentways. Beneficial to society it’s all going to come down to understanding if wewant to accept our bias, our own bias, and I think this is going to be thehardest part for people and there is plenty of examples as recent examples ofover the last few years of just not understanding the data that we have, the processes we have in place that create bias, we just didn’t know we neverlooked at the data so I am always hesitant to talk about beneficial impact tosociety because it might have a negative impact first, overall I think it’stotally interesting as to what we can now achieve and if we can start toreplace some mundane tasks can I now do something else. Can I start a whole newart renaissance, right, with this concept of a 4th Industrial Revolution? Butif I go back to those sensing capabilities it’s not them individually its themcombined as a whole because now if I can start to, I don’t need to necessarilyhave sensors to detect on a manufacturing line if something going wrong, if wecan now sense with smelling sensors in some way shape or form thatsomething is starting to burn, a conveyor belt, or a rota or something you know,with vision again going back to manufacturing can I detect that something is alittle bit out of skew in a car or as its being built or that door being laid downin a perfect position there are all sorts of things that we can do but I thinkwe have to understand how can we bring the human back into that, the human nowunderstands the sensors and can they take what that system is telling it and dosomething to help.
Then again just as a side note a little bit of ajoke, but if you look right now there are so many car recalls happening. I meanin the millions of them, all the manufacturers. Could we start to detect someof these problems if we had a little bit better sensors, going down rightinside of their testing and inside of their manufacturing. So again, itwas just I read something this morning on the vast number of them growing,right, and you could be beneficial that way.
But you could also be beneficial in helping someonesee if now we can start being in some of the transplanting chips into people.We have all sorts of things we could do but I also know there is going to be alot of negative impact and I think we have to balance that and understand thatthe way it is, I learn from it, moving forward.
Rob Telson: Yes, it’s a broad horizon out there, Glad youbrought up the point on industrial and manufacturing and the impact thatyou can have throughout the whole supply chain and how there can be thesepositive attributes for being beneficial no pun intended but you are right theoutlook on AI in almost every, it does have that social media feel to it.I remember when I logged onto FaceBook which was eons ago, put up some picturesof my family right, and look at social media today, it’s got a lot ofpositive attributes, but it’s got a lot of challenges to it aswell. As we evolve and Ai is going to have some of that and Ilike the fact that you brought up transplantable devices and how we canhelp people better, or touch or sense, I should say more effectively.
There is a lot of dynamics to it, we will takeit one step at a time, but I think that as we see it evolve there is a lotof good things that are going to come out of it, as long as we getthat human layer dialled in. So, Rob with cybercrime continuing to growthere are new techniques in cyber security which are Introduced tomitigate the increased levels of risk this is an area where you have beeninvolved in analysing and educating the industry being able to process onthe device without the dependency on the cloud can provide another layerof device security. What other areas do you feel Edge based Ai processorscan make a positive impact on device security?
Rob Lincourt: I think there’s a lot right, I mean if you thinkabout it today were throwing in, and you made the reference to thesebillions of IOT devices, into the environment and it’s growing evenmore rapidly.
Think of all the sensing you now have in a car. Ifyou look at some of the car platforms they actually have cameras on thedriver to detect if the driver is paying attention and how they interactis a separate issue but imagine if somebody could take control of that, andimagine now if that had to travel all the way back through a network, evenif it’s a 5G network, all the way back to a public cloud, get routedall through the public cloud firewalls, all the way back to yoursystem. Your service has to call 3 or 4 other services, includingthe service that does the object detection and, and kind of an understanding aview of my face and my eyes to determine if I am distracted and then come backthat could take even if it’s a second or two seconds if you are travelling at80 mph on the highway that’s still a pretty good distance right, so you have toworry about that, now the car itself can protect it using other sensors, butagain what if someone got into that car, can anyone now push code down to yourcar and execute it. You’ve got to be very cautious with how that movesaround. You have to start building out this concept of trust.
Trust in compute environment all the way from thecar, all the way back to the cloud and everything else but it I can push modelsdown onto the car that have been trained and up dated with different weightsand coefficients and to not send back true data with coefficients, you couldtweak, but if they get out of range you just don’t have the system acceptanceso there’s other ways you can do it, you know even going back to the camera andthe doorbell, if you now can lean my face do you have to send my image back tothe cloud or can you send weights and coefficients or kind of data that makesup my face whether that’s point cloud, however, whatever type of camera youwant to use there are so many different ways of doing it that doesn’t mean youdon’t go into how Ai now can be imbedded in infrastructure to actually look forand you know detect anomalies, detect unauthorised access from differentpattern uses and even the way I type on my keyboard if something’s wrongsomething like that so there are a lot of things you can do in that space soits limitless right now.
RobTelson: Itotally agree with you. You brought up right there at the end embedding Ai andchecking patterns out and looking at behaviours it’s going to play a key rolein security through the ecosystem as we continue to evolve.
The last couple, if you could be a super hero whowould it be and have one Ai super power what would it be?
RobLincourt: Answer not transcribed
Pod castconcludes with thankyou to Rob Lincourt for participating and confirmation thatthe next podcast in the series will be next month.
How do you see neuromorphic processingchanging the face of Ai?
(20min delay)