Silicon Valley Innovation Center Interview Logo

INNOVATION WITH MACHINE LEARNING

Interview with Julian Nolan, Founder and CEO, Iprova

ABOUT THE INTERVIEW

In business today a key challenge is to generate not just new ideas but new ideas that have real commercial value. But where do these ideas come from? Should we look to make incremental improvements on already established products or is it better to dive into completely new domains and start from scratch? By studying trends and technologies in markets across the world, can we jumpstart our own innovation process?

To answer those questions and more we’ll be joined by Julian Nolan, the founder and CEO of Iprova. Since its founding in 2010, Iprova has been coming up with new inventions by using machine learning to analyse real-time data from a variety of seemingly unrelated fields. Julian will share why the emergence of machine learning challenges traditional notions of what it means to innovate.

Watch our Facebook Live interview here at 9:00 am PDT / 4:00 pm GMT on May 21. No registration required.

ABOUT OUR GUEST

Julian_Nolan_Iprova

Julian Nolan

Before starting Iprova, a data-driven invention company, Julian was Vice President of Licensing for Honeywell in Europe. Formerly, Julian worked at the Central Research Laboratories of EMI Music in London in the role of business development director of the DSP group. Julian holds a first degree in electronic engineering, and postgraduate degrees in subjects that include Artificial Intelligence.

INTERVIEW TRANSCRIPT

Rahim Rahemtulla:
Hello and welcome to this Silicon Valley Innovation Center interview. I am your host, Rahim Rahemtulla. And joining me today is my guest, Julian Nolan. He is the founder and CEO at Iprova. And so, Julian, welcome to the program, thank you so much for joining us. The title of our talk today is “Innovation with Machine Learning.” And that is for good reason because your company, Iprova, is, of course, involved in this field, just to give the audience a little bit of context there. So let’s just dive right in. We have innovation, we have machine learning, let’s bring them together. What does machine learning bring to the innovation process? How does it work?


Julian Nolan:
Yeah, well, thanks, Rahim, it’s a pleasure to be here. And thanks for the invite, I really appreciate that. And so I’d like to take a step back and look at how inventions and innovation traditionally occurs. So, if you think about it, very often people might go to an exhibition, they might go to a conference, or they might meet someone for coffee, and they learn something new, they apply that new thing in their area and they have this magic inventive moment and come up with a great idea. Now, if you think about it, it’s actually very suboptimal. Who’s to say they couldn’t have gone to a conference at an earlier time and done something earlier that would have led to the same idea? The one way it’s suboptimal is in timing. Now, who’s to say they couldn’t have met someone else for coffee and find out something even better and come up with an even better invention? So it’s also suboptimal in terms of what we call “invention diversity” or the disruptiveness or the commercial value of the resulting inventions.

And so what we’re doing is to really bring together the right information at the right time to allow people to invent, to innovate in a highly optimal way which delivers very commercially relevant results. And so the way we’re using technology is to really understand, on a massive scale, those advances that are going on around the world at any particular moment in time and then to be able to relate those to the specific area where we’re creating inventions. Now, it has to be said, what we’re doing is to enable humans to invent faster and better than ever before; what we’re not doing is to create the invention using a computer. So our technology, our machine learning, our data feeds the right information at the right time to our human inventors so they can invent in a highly optimal way.

Rahim Rahemtulla:
So you talk about having coffee with someone, going to a conference, whatever it may be where that spark comes from. So does the technology essentially allow you to simulate that many, many times over? Because you can’t go to all the conferences, you can’t meet all the people for all the different types of coffee, but as if you could multiply yourself and do that hundreds of thousands of times, you increase your chances many times over.

Julian Nolan:
Yeah. Exactly right. So, of course, the level of information in the world, more and more conferences, as we all know, essentially more and more people meet for coffee. After all, a lot of companies are growing at the moment. And simply it’s impossible. There’s too much information. There’s no way that any one person has of knowing all of the advances around the world that may be relevant to their area, particularly as today we have huge convergence between previously unrelated areas. So, for example, many of the inventions that we may rely on an advance in, let’s say, human biology, triggering an invention in, for example, autonomous vehicles. You have industries that have previously been unrelated coming together for the first time. And, of course, it’s simply impossible for someone who’s, let’s say, working in eponymous vehicles to attend all of the conferences, look at all of the applications that might trigger an invention in their area. And that is exactly what we’re doing with our technology.

Rahim Rahemtulla:
And I wonder, Julian, it’s one thing to have the ideas or have the raw data, so to speak, or get the inspiration from maybe another industry or something that you might have seen elsewhere. But converting that then into something that you can practically use or sell – that’s a whole other kettle of fish, so to speak. That’s, I think, a lot of the time where big corporations, for example, have trouble – converting the ideas into something that they can really use. So what is the technology – machine learning, for example – what can that tell us about that problem? Is there a way around that? What are your thoughts there?

Julian Nolan:
Yeah, so that’s a great question. And what we’re focused on is creating highly disruptive inventions. Typically, our customers are very well able to create great inventions that stem from advances in their own area. So, for example, let’s say you’re an autonomous vehicle company, you would know all about the advances in autonomous vehicles and you’re well able to create advances in that field which relies on advances in that field. Now, very often a lot of the commercial value is in being first to recognize a point to a convergence between one industry and another. Let’s say, between autonomous vehicles and healthcare, for example. And that’s exactly what we do. So what we’re doing with our technology is to be able to create an information asymmetry in our favor so we can recognize these points of convergence at a very early stage and create the foundational intellectual property for our customers in those areas. And, of course, identifying things at an early stage means that you are in a great position, then, to take commercial advantage of the resulting opportunities. And because of that the inventions have intrinsic value to our customers.

Rahim Rahemtulla:
And so what happens, then, when you have this? Do you build a prototype which you then will go and test and then sort of iterate on it, for example? I mean, I know that’s a popular method of doing innovation.

Julian Nolan:
Yeah, so we don’t build prototypes. So our business is solely focused on creating high-value inventions. So we document the inventions and we deliver the inventions as a stream to our customers and our customers are then free to accept or reject inventions from the stream at their entire discretion. But what we don’t do is construct the prototypes. What we do is we demonstrate the invention works by citing things that have gone before the invention and existing technologies.

Rahim Rahemtulla:
So I just want to circle back to talk a little bit more about machine learning, which I guess, it’s fair to say, has been around for a while in some ways but in other ways it feels like it’s only just emerging and there’s a lot more applications of it now than there have been. But, of course, you emphasized earlier that what you’re doing there does eventually come down to the human inventors who are just using that data. So I suppose my question is where you feel the line is between machine learning and humans and, as we go forward, trying to decide where we use artificial intelligence, where we use human intelligence? And there’s value judgments, which in your line of work as well must be made at some point. How much do you rely on the technology to do that? How much do you rely, at the end of the day, on human creativity and ingenuity?

Julian Nolan:
Yeah, so that’s an interesting question. And what we’re doing is all about ultimately augmenting and enhancing the creativity of the user. So what we want to do is to allow our human inventors to invent in this highly optimal way by enhancing their ability to be creative. And, of course, we want to do that as much as we can. One way of thinking about what we do today is a bit like cooking: the software is bringing together one or more of the ingredients that are required for an invention, then our human inventors assemble those ingredients to create the invention. Now what the software is able to do is to bring together, effectively, the right ingredients at the right time to create dishes that are highly relevant or very tasty, but which have never been seen before. And so what we’re looking to do as a company, of course, is to really allow people to be more creative than they would be otherwise. And I guess that means two things. It means making people who are already creative yet more creative, able to create the right invention at the right time by having access to the right information at the right time. But it also means making those people who might not be creative, allowing them to be perhaps creative as well. So it’s really almost a democratization, if you like, of creativity, allowing people to come up with inventions which otherwise they would be unable to do, at least in a time-optimal way.

Rahim Rahemtulla:
Sure. And so can you give us an example? What are the sort of ingredients that you speak of? To give us a better picture of what we’re dealing with here. What are the sources? Is it information? Is it numbers or more quantitative rather than qualitative? What are these different elements that come together?

Julian Nolan:
Yeah, so the kinds of things we’re interested in are actually things that are surprising and unexpected in a given field. I can give you an example. So something we made a few inventions on some time ago is the fact that people have light-sensitive proteins in their ear, that if you shine a light in your ear, it has the same biological effect as observing light through your eyes. And this is something that can influence your circadian rhythm and your body clock, that has a lot of application in, for example, the lighting area. Now, for us, the fact that humans – in fact, all mammals – have light-sensitive proteins in their ears, that is surprising and is unexpected, particularly if you apply that outside of the domain of human biology. But that’s an example of one ingredient that we would use.

Rahim Rahemtulla:
And so that would then be of use, you say, to the lighting industry, for example?

Julian Nolan:
Exactly.

Rahim Rahemtulla:
Or many, many.

Julian Nolan:
It could be a number of other industries, but what we’re doing is to really track huge numbers of domains, the advances that occur within those domains, and then understand if they can trigger inventions in other, perhaps distant, areas. And the kind of qualities that we look for are things that I would say are surprising and unexpected, which is, in our view, one of the main constituents of things that people perceive to be creative.

Rahim Rahemtulla:
And then, yes, this element of surprise and something unexpected is really interesting, I think. It sort of adds this element of mystery, almost, to the process, which I guess could make some companies uneasy if they don’t know where that’s going. But I suppose what you’re saying is you take that element of surprise, but then you actually apply it somewhere and turn it into something that is going to have commercial value.

Julian Nolan:
Exactly right. You are exactly right about that. And we also document the invention in such a way that it’s very transparent. And I guess what should happen is that the invention should be surprising and unexpected but yet, of course, very commercially relevant and not at all mysterious. So it should be something that our customers can readily understand and certainly not be mystified by it. It’s part of our job to explain it in such a way that it can be readily understood. And the best inventions are normally the simplest ones.

Rahim Rahemtulla:
And that’s interesting. So tell us more about this. So is that difficult to do sometimes? You must sometimes find really amazing, unexpected things which are sort of almost begging for a practical application.

Julian Nolan:
Yeah, that’s very true. So we work with a lot of leading technology companies around the world and I think different companies have a different time horizon for when they’d like to bring an invention to market. So what would make a great invention for one company might not be good for another simply because one company might want to commercialize an invention within three to five years, but another company might be focused on longer term, maybe eight to twelve years’ commercialization. So according to the time horizon that our customers have when they want to bring the inventions to market, we’re able to tune the inventions to match that based upon the ingredients that go into making up the invention.

Rahim Rahemtulla:
So how well do, when you do talk to your clients, how well do they understand this innovation process? Because, of course, today we do hear a lot about how all companies really need to be able to innovate, more so now, perhaps, than they did previously because of the advances in technology, whereas for you guys, innovation is your business. So, of course, you’re always willing to take that mysterious journey, you’re willing to look far and wide and you understand that that’s not always going to lead to something with commercial application but, of course, you’re going to go there anyway and you’re going to find out and you’re going to go through that process. As someone who talks to these bigger companies, how well do they understand that part of innovation is simply having to make that journey? You don’t always know what’s going to come out of it, there is a lot of uncertainty there, but it needs to be done.

Julian Nolan:
Yeah, that’s true. And, of course, it’s impossible to know the future. At the same time, it is possible to have a good guess and an informed guess as to what the future might look like. And we would never claim that we know the future. Of course, that’s not true. But, in a way, it’s our job to put inventions to our customers that represent their view and our view of what the future might look like based upon points of convergence that are very, very likely to occur. Again, if you take something like healthcare and autonomous vehicles, the inventions in that area only makes sense if you believe in autonomous vehicles, which, of course, a lot of companies do, and you believe in a growing market for health care, which, of course, a lot of companies do. But bring those two together, that’s something, then, that could be quite interesting for companies in the healthcare area or the autonomous vehicle area, even though today healthcare in autonomous vehicles isn’t something that’s deployed.

Rahim Rahemtulla:
No. But you guys could bring that together? Because you make the case for that as well because, obviously, we can say “Healthcare, autonomous vehicle. Okay, that makes sense,” but to actually get a company to invest in that idea, they need something more, something a bit more robust than that. But presumably, you do provide that. Or what do you provide in that case to make that case, to make them feel that the risk is, in some sense, justified?

Julian Nolan:
Exactly. So what we do is we give them patentable inventions, so allowing them to create foundational IP in that or indeed many other areas. And of course, these inventions are very specific. So it’s not just saying, “Hey, there’s a point of convergence between healthcare and autonomous vehicles.” There’s tons of companies that can do that. We’re giving them very specific technical inventions that can support a patent application. And so if you think about this area, one of the inventions that we’ve created is where we use the motion of the vehicle and the fact that you can control the motion of an autonomous vehicle in a very accurate way and also then use a camera – which of course you have an autonomous vehicle already – to detect the response of a passenger to that motion. Based on the response of the passenger to the motion of the autonomous vehicle, it is possible to diagnose medical conditions that depend on your core balance. For example, diabetes and multiple sclerosis. So then, of course, you have a way of having a medical checkup for certain medical conditions as you take your AV on the journey to work, for example, or if you go shopping. So it’s a great way of utilizing the time that you’re spending in a vehicle that otherwise might not be able to get used.

Rahim Rahemtulla:
That’s kind of amazing actually, to hear you describe that. Because, I don’t know, maybe it’s obvious to certain people, I just think, how would you ever have come to that idea? How could you ever have thought, “Oh, maybe there’s a crossover here. Maybe this will work or we can use autonomous vehicles in this way.” I mean, that’s just not something that comes to you. So this must be what you’re talking about, this is the process. This is how you hit upon ideas like this, which just seem not necessarily counterintuitive, but just beyond what any individual might dream of or dream up.

Julian Nolan:
Yeah, exactly right. So it’s our job to come up with inventions that are additive and complementary to those that our customers can come up with themselves. Our customers are normally great at coming up with inventions based on advances in their own area. But, of course, if you’re an AV company, you might not know about advances in healthcare or in human biology. But we do, and we can relate those and understand which of those advances can trigger inventions in, for example, the AV area or indeed any area. So this is then how we have the basis and the information with which to come up with these inventions. Because in some ways we look at the human brain a bit like a transfer function – what you get out as a product of the information that you put in. And so what we’re trying to do is put in the right information so we can get the right inventive outcomes.

Rahim Rahemtulla:
Yeah, okay. Julian, we’re almost out of time, unfortunately. And maybe just to bring everything together then and maybe draw a focus just a little bit on the role of the machine learning there as well. I suppose we haven’t perhaps mentioned that as much as we could have. Is machine learning a game changer here, in some sense? Is it that it just gives you more information to work with at the end of the day? Or is it offering sort of a fundamentally new product, if you will, or a new approach to innovation.

Julian Nolan:
So, for us, it’s absolutely core to the technology that we have. And we, without using the machine-learning techniques and without having access to the training data that we’ve developed, which is proprietary to Iprova, we wouldn’t be in a position to have the technology that we have today, nor provide the information that we do to our human inventors. So it’s really at the heart of what we do. And I think that the reason for that is that the problem we’re trying to solve is extremely difficult because we’re trying to find things that are extremely distant or quite possibly very distant in area, but understand if these small inventive signals can actually trigger an invention in, potentially, an unrelated area. So you have very, very small signals and certainly whilst you can try and solve the problem with a number of technologies, certainly machine learning, with the right training data is, in our experience, the only one that can really provide the solution.

Rahim Rahemtulla:
And then, if we sort of step back from the technology a moment, is the message then for all the innovators, the entrepreneurs, the intrapreneurs who are out there, is the message then that you really have to have a very wide field of vision? That the further you look and the more distant an idea may seem, I mean, does that in fact give it more potential to be that disruptive invention or that disruptive breakthrough?

Julian Nolan:
Yeah, that’s really true. Because in a way, in a sense, people are burdened by their knowledge. So it’s no longer enough to be a computer scientist, you have to be a specialist, let’s say, in natural language processing. And of course, if you’re a specialist in natural language processing, you spend all your time focused on that topic, so it’s very difficult for you to know about advances in, let’s say, telecommunications that could trigger a great invention in your area. And so I think that’s what we’re trying to do. It’s basically give people this wide angle view of the world which complements very often the telescopic view that they necessarily have because of their knowledge and deep level of knowledge in a single area.

Rahim Rahemtulla:
Wonderful. Well, Julian, thanks so much for joining us today. It’s been a pleasure talking to you, hearing about innovation and how it is working with machine learning. I think that’s where we’ll have to wrap things up today. So just a big thank you to you for joining us on the program.

Julian Nolan:
Thanks for the invite. Thank you very much.

Rahim Rahemtulla:
And to our audience, thank you as well for tuning in today to hear my conversation with Julian Nolan, the founder and CEO at Iprova. I’m afraid that’s where we’ll have to wrap things up for today, but thank you very much for watching. Do join us again for the next interview. See our website siliconvalley.center for all the details on our upcoming events. Until next time, goodbye.

info@svicenter.com 1-650-274-0214
[FirstName]
[FirstName]
[LastName]
[LastName]
[Title]
[Title]
[Company]
[Company]
[email]
[email]
[w-.]
[w-.]
[w-]
[w-]
[FirstName]
[FirstName]
[LastName]
[LastName]
[Title]
[Title]
[Company]
[Company]
[email]
[email]
[w-.]
[w-.]
[w-]
[w-]
[FirstName]
[FirstName]
[LastName]
[LastName]
[Title]
[Title]
[Company]
[Company]
[email]
[email]
[w-.]
[w-.]
[w-]
[w-]
[FirstName]
[FirstName]
[LastName]
[LastName]
[Title]
[Title]
[Company]
[Company]
[email]
[email]
[w-.]
[w-.]
[w-]
[w-]
[FirstName]
[FirstName]
[LastName]
[LastName]
[Title]
[Title]
[Company]
[Company]
[email]
[email]
[w-.]
[w-.]
[w-]
[w-]
[FirstName]
[FirstName]
[LastName]
[LastName]
[Title]
[Title]
[Company]
[Company]
[email]
[email]
[w-.]
[w-.]
[w-]
[w-]