Rene - Hi, Welcome to QuBites, your bite-sized pieces of quantum computing. My name is Rene from Valorem Reply and today we're going to talk about enterprise quantum computing and quantum machine learning again and for this I'm very honored to have a special expert guest today, Santanu Ganguly. Hi Santanu and welcome to the show. How are you today?
Santanu - Hi Rene, thank you very much for having me on the show. I'm pretty good, I hope you are too, during these dynamic times. So yeah, thank you for having me here.
Rene – Awesome. So before we start with the questions, can you tell us a little bit about yourself and your background as it relates to quantum computing, computer science and you know all the field here.
Santanu – Sure! I was always a student of physics and in fact I was doing my PhD in UK and in physics and then I got an opportunity to work in Switzerland in network engineering, sort of, you know domain. So I thought, hey, it'll be nice to go and live in Zurich for a while, So I left my PhD and went to work in Switzerland and I enjoyed it very much. But fundamentally my background is in Physics and Math. I've got a postgrad in both subjects and then of course I spent quite a few years in computer science and related subject matters as well. So that's my basic background. That's who I am, thank you.
Rene - Awesome, and so you are also working at Cisco, and which is a large enterprise company, famous for networking and many other things but what are some of the topics at Cisco, when it comes to quantum computing and what role is quantum computing actually playing in these kind of enterprise systems at the moment?
Santanu - That's a very good question. So to answer the first part of your question, it is absolutely right, Cisco is basically in 70% enterprises across the world and we've got a very healthy customer base there. Cisco has traditionally led networking servers, computer systems software everything in the industry and we are also aware that quantum computing and security related to that, is something that is upcoming, something that is happening today as per NIST, as per NCSC and we are actually very consciously in there, working with the standards divisions to meet those criteria. So that's where we are. We are telling the industry that we have our foot in there. We are working with post quantum crypto currently. So that's our model right now. We are also looking at other branches of quantum computing. Not just quantum computing but also communication. For example, today there are different types of quantum computers. You have super conducting computers from IBM or Google. You have annealing based computers quantum computers from D-Wave, you have photonic quantum computers from Xanadu, but none of these computers, even though they are the same hardware, has ever communicated between each other and this is where Cisco comes in, because we are a communication company. So that is our major focus in quantum internet. Now a lot of people will laugh at us or me if I say quantum internet and it is true. it's quite a few years away but if we sit back and tell people it's quite a few years away and we won't work on it, that's not gonna get us to where we wish to go. So our policy is to follow Abraham Lincoln saying that, you know, you need to build your future, to ensure your future, you need to build it yourself. So that's where you are. Thank you.
Rene - I love this statement you were just saying right it's like you can always say, oh yeah that's too far away, it's like oh that doesn't work. But if you never move, if there's no early mover, no one will move right? So someone got to take and it takes brave people like you and the team to actually do that and you know work on it on a mission and also you know, myself, of course, I'm also just working on emerging technologies and some of those might not turn out to be a huge business but all of us will and that's, you know, you got to take that and you got to work on it especially like you were saying Cisco being a large network company and maybe another question regarding to that the whole QKD, like the quantum key distribution and the amazing progress. I just saw the other day that in Japan they made a new record in June this year, I think, with a 6 KM long quantum key distribution channel, which is quite impressive, but I guess that's also something you guys are looking at right?
Santanu - Very much so. So currently Cisco's vision and we, as we speak, we have a solution as in our software implemented for post quantum cryptography. When I say post quantum cryptography, I mean using classical encryption not quantum, but classical encryption that is theoretically supposed to be quantum proof. So why are you doing this because again. As I said we are following the standardization but now that we have a PQC solution, we are also working with universities and research institutes to look at options of taking it to the next step. So when if and when the standards body decides that, okay, today is the day that QKD should go into networks, we will be ready if that day ever happens and at the same time QKD is also part of that communication stack that I spoke about. When you have two quantum computers communicating between each other there are stacks of security, there are stacks of Communication, there are stacks of resource management, various stacks and that's something we need to think about, not just transforming one qubit to from one point to another and then you know because that does not serve a large scale purpose. So yes, you're very correct, we're definitely looking at that.
Rene – Yeah, so the scalability is of course the challenge, but this is important to actually move it into production out of the research labs and that's one of the pieces you guys are working on. Awesome. Is there any other thing you would say where you're already seeing impactful applied quantum computing solutions like quantum inspired computing or anything else where you would love to share an example about.
Santanu - Sure. I think major impact right now as we speak, one of the major impacts is in chemistry and molecular modelling. And that's not just quantum computing, there's also an aspect of quantum machine learning there. So, you ask for an example and I think one of the one of the ready examples in my head right now is the research team at Penn state university in the US who are actually looking at modeling of covid vaccine using quantum computing and this is a research group being led by Dr. Swarup Ghosh and they are they actually started working on it I think early last year. So that's one example of it but there are various other examples in drugs industry, in protein folding and that's just that's just in chemistry and molecular modelling. On the other hand, there is a very large push from the financial industry towards quantum algorithms and machine learning as well. So, you may be aware that Goldman Sachs has built up a quantum computing research team, JP Morgan is doing the same thing and all these guys are doing it because they see an advantage there that they have missed or probably they think they cannot get in classical computing.
Rene – Right. It's this, I mean, if we think about portfolio you know risk analysers and all of these problems these are these super like NP complex problems like that are just you know incredibly hard because they're exponential and with classical computing it doesn't work that well, but with quantum computers you basically add another qubit and you get another exponent at it, so you can solve these exponential problem and well linear polynomial time depends really on the problem but it totally makes sense. Actually in fact at a quantum conference earlier this year, I don't remember which one, but there was a speaker also talking about quantum finance and he basically said like financial institutes need to get on the wagon now, because otherwise if they miss the opportunity they will be gone out of business in like five to ten years, right, because they cannot, their competition will be so strong basically with all the financial institute that invested already in these kind of technologies. But speaking about quantum machine learning, I also know that you wrote a book about applied quantum machine learning which sounds actually super exciting. So, can you tell us a little bit more about your book? First of all, the title and where folks can get it but also what you cover in there, we'd love to learn a little bit about it.
Santanu – Sure. The title of the book is quantum machine learning and applied approach and I went for the applied approach because prior to starting to write the book, which I did in during the lockdowns last year, there were two I think mainstream major works on quantum computing one was the first ever one was by late great Dr. Peter Wittek. He wrote a book on quantum machine learning back in 2011, I think. Then there's the seminal volume by Dr. Maria Schuld and Petruccione, Supervised learnings with quantum computers, that came out in 2018 and they were great read, especially Maria Schuld’s book, I mean, really enjoyed it. I have it right there and the thing is, I enjoyed those books but what I realized after I went through them and after I started to you know dabble in hands on quantum computing itself is, like I said before, there are so many various quantum platforms there IBM has one, Google has one, Xanadu has one, D-Wave has one, Rigetti has one and several others. Now you have these whole bunch of quantum algorithms the theory of which are all explained in these books very nicely. How do you apply those theories and those algorithms to these diverse platforms, number one. Number two, do all those algorithms give you similar advantage on every different platform, so these are these are things that sort of triggered my interest in writing a book with an applied approach. So my approach or what I tried to do is introduce the theory and then introduce the coding bit, be that google, be that IBM, be that Riggeti. So I'm very grateful to DF systems they actually reviewed my two chapters on DF systems to go through where you mentioned the np heart problems. I actually did a code a sample code on traveling salesman problem and then and regetti max cut problem which are all NP hard so that was my thinking that look there are lots of people like me around who have all these four or five different platforms in front of them and they have all these theory which they understand but perhaps it'd be good to try and bridge that gap where they understand and where they apply so that's where it came from.
Rene - That's amazing and exactly I think these are really the missing pieces here to, you know, help people to actually get into quantum. Because you have a lot of academic and scientific like papers and books and what not but like this applied approach that you cover in your book this is really missing in the market. So folks get a copy, I will also order one of course. It's definitely worth investing because this is definitely the future and maybe the last question for you regarding QML and in general what I was just talking recently with Johannes Oberreuter in another episode of QuBites which you should all check out about QML in general and the image classification, QML image classification it was telling me like the only results they have even with small quantum computers basically you inject a layer in your neural network with for quantum computer and a later stage right when the features were basically extracted and they were seeing like better accuracy already even with these small quantum computers, why is that the case?
Santanu - I'll try to explain my point of view on that. So image processing as you correctly point out, is a very important part not just not just for academic purpose but also for security and the reason for that is if and when we are ready to have quantum sensors out there, some of which are, being produced right now, what you'll get is, you'll get 3D data, data in three dimension, which is essentially quantum data. Once you get quantum data, so okay, I'll first try to answer your question and then I'll go to what I'm saying. So your question was existing data and why some results are better. The reason for that is again, when you take a classical image and you piece it up, you chop it up in little pieces and you transfer it into a quantum image processing algorithm, there are several, NEQR comes to mind, FRQI comes to mind, what it does is it gives you that probabilistic application of quantum mechanics to optimize or deteriorate your image further or optimize it further or deteriorate it further. I'm not saying every run of those processing or machine learning runs will give you an amazing optimized image, it may not, but some runs will give you that's optimize better version, the reason is, you take, when you transfer when you process a classical image into a quantum machine learning process, you take a classical image and what we typically do is we do feature mapping. So from a classical XY 2D image we map it into a hilbert space where it's more than two dimensions and is those is those multiple dimensions that gives it the added power because you get so many various different probabilities to apply. And this same thing will apply once we get what i was talking about a 3D image or quantum image and then we should be able to see around the corner. So if you have a building, we should theoretically be able to see what's happening behind the building once we have that kind of data and quantum processing.
Rene - That actually makes a lot of sense. You can represent your data better and closer to the nature if you will. So that's true, yeah, that's awesome. Well, unfortunately we're already at the end of the show. I think we could talk for many more hours. Thank you so much Santanu for joining us today and sharing your insights. That is highly appreciated.
Santanu - Thank you for having me and well thank you very much for your kind words and thank you for your attention.
Rene - All right folks, thanks everyone for joining us for another episode of QuBites, your bite-sized pieces of quantum computing. Watch our blog, follow all social media channels to hear all about the next episodes and when we release them. So far, take care and see you soon! Bye-bye.