1165 København K
Tlf: 35 32 28 98 (mon-thurs)
Machine learning is the secret ingredient behind many web-based services and applications that we use every day. Corinna Cortes, Copenhagen professor, head of Google's New York research centre and developer of one of the main algorithms in the field, talks to the University Post about her work, her time, and her running
Google products are based on complex algorithms that train computers to recognise differences and similarities between objects.
The woman heading some of the most advanced web-based services, like Google Voice Search, is Corinna Cortes. She works at the Google research center in New York, and is at the same time a professor at the Department of Computer Science, University of Copenhagen.
The University Post interviewed Cortes about her work at Google, her role as a professor and her passion for running.
You have a Master’s in physics and a PhD in computer science. How did you move from one field to the other?
»During my studies in physics I did a lot of computer simulations, and my supervisor said ‘well, you should get a PhD and maybe computer science is the way to go’… «
Do you miss physics?
»I don’t miss physics now, maybe also because I don’t have enough hours per day… My life is so full with challenges that I don’t have time to miss anything.«
Could you describe what your research work consists of?
»I’m doing mainly machine learning. In particular, together with my team, I am searching for what is the best method to teach computers how to find similarities between objects, that could be words, images or whatever. In brief, what we look for, is the position of the objects on a curve, called the energy function, that quantifies the distance the objects have from each other.«
»Together with the members of my group I try to develop good algorithms. There are three characteristics that make an algorithm good: First of all, it has to work not only with the examples used to train the computer but also with unknown objects. Secondly, the algorithm has to be efficient and to solve problems fast. Lastly, I like to develop algorithms built on a solid mathematical basis, so that I can explain why my algorithm works fine. And this is why my code is better than the one of the guys in the office next door!«
Machine learning can be applied to many practical problems. Is the practical problem you are working on more important to you, or the way you find a solution?
»I think that whatever you do has to be challenging… What is important is that if I come up with a good solution I can touch a lot of people. This is the motivation that keeps me working.«
Where do you get your inspiration?
»What I do is research oriented, but the starting point is always a real need for the company. When I have to find something new to work on, I usually go take a cup of coffee. Along the way to the coffee machine I bump into the software engineers working close to me, and we start talking on what we are working on. This is the way we use in my group to find interesting projects.«
»The goal of my research group is to help a specific engineering team, and hopefully come up with a general solution that can be used also by other teams. Typically, we develop open source libraries that can be used in a number of different situations. For examples, our libraries are used to run several speech applications at Google, like Voice Search.«
What is the difference between research at Google and research in academia?
»The starting point is different. At AT&T I was working on projects to be used by the company. At Google we are also close to the product developing process, but at the same time we have the freedom to continue working on theoretical problems.«
»Usually when students ask ‘why should I remain in academia?’, professors say, ‘because here you can have all the freedom of the world! If you go to a company, they will tell you what to work on’. But when I am on panels, a professor came and asked me ‘what are the interesting problems you are working on?’. And when they have the problem, they wonder where to find some data to work on.«
»At Google we have interesting problems, we have the data, and we do real theoretical work. So we are the best of all world and I think students should come to us when they are looking for interesting problems!«
It seems Google offers all you need if you are a computer scientist. What motivates you beyond this choice to invest time in academic research also?
»I would like to guide students towards problems that are relevant for large scale companies like Google. And since the professors come asking if we have interesting problems to work on, I have decided to help them directly. Also, it is important for us to be present in universities.«
Why did you choose to be a professor at your alma mater, the University of Copenhagen?
»It was natural for me to choose something in Denmark, because I’m Danish and I like to have connections with Denmark. It is a very good university and I’m definitely proud of coming from there.«
You are head of a research centre, professor, have two kids and are a competitive runner. How do you manage your time?
»My golden rule is to do something every day, preferably in the morning so at the end of the day I’m always sure I have done something. Besides that, I’m terrible at time management: I do plans but then everything goes crazy and at the end of the day I don’t even know my middle name. And I don’t have a middle name, by the way. I work on weekends.«
Let’s talk about your passion for running. How often do you run?
»I try to run every day.«
Do you use any gadget to train or you prefer to run free?
»I’m very old fashioned: when I go running I just bring my phone and keys with me in case something happens. My phone is an old one, pretty much without a display but almost waterproof, if you leave it for a while on a radiator after a heavy rain. But I like to plan my track using a website called Gmap Pedometer so to know how I far I run. And when I’m in a new place I use it to see where to go.«
Read a 2014 interview with Corinna Cortes on the subject of Big Data here.
Like us on Facebook for features, guides and tips on upcoming events. Follow us on Twitter for links to other Copenhagen academia news stories. Sign up for the University Post weekly newsletter here, and then follow the University Post on Instagram here.