Nicu Sebe is a professor in Computer Science at the University of Trento, Italy, where he was the director of the Department of Information Engineering and Computer Science.
Back in 2005, an interesting result was collecting dust on Nicu Sebe’s desk, who, at the time, was a researcher at the University of Amsterdam in the Netherlands. Little did the rest of the world know that the Machine Learning specialist back then, and the current Head of Artificial Intelligence at Humans.ai, was holding beside his steaming cup of tea the key to the mystery behind the world’s most famous smile, the one portrayed in Leonardo Davinci’s Mona Lisa. This all changed when a journalist stumbled across this research, and soon after, Nicu Sebe’s name circled the entire Globe.
The Romanian professor designed a computer software that deciphered Mona Lisa’s enigmatic smile, revealing that it was 83 per cent happy, 9 per cent disgusted, 6 per cent fearful and 2 per cent angry.
Sebe’s solution to cracking the Mona Lisa was straightforward and elegant in its approach. Developed as a joint venture between the University of Amsterdam and the University of Illinois at Urbana-Champaign, the software works by examining key facial features and neutral expressions of female faces loaded into a database and comparing them against a reference image (the painting in this case).
How did Nicu Sebe become one of the leading figures in the field of Artificial Intelligence, and what is the story, and motivation that determined him to join Humans.ai as the Head of the AI department are all questions that will be answered in the first interview from the Meet our #SuperHumans series.
We sat down with Nicu Sebe on a sunny day in January at the Humans.ai office from Bucharest to chat and learn more about the choices that helped put the talented Machine Learning professor on an incredible career path and how he ended up leading the AI developing process at Humans.ai.
Can you tell us a few words about yourself, what is your professional training, what pushed you to work in this very advanced technological area? What was the trigger that led you to AI?
Well, it’s a long story […] It’s nice to remember it sometimes. I studied in Bucharest, I did the Faculty of Electronics, and then it was like a challenge to try to go abroad. I think it’s a dream that many people share.
I applied at several universities, mostly in Belgium and in Holland, and to my surprise, I ended up having several offers to do a PhD, and somehow, I chose the University of Leiden from the Netherlands. I started there in 1997. Then I had, I would say, the opportunity of my life to meet professor Thomas Huang who invited me to go and join his group. So, in the end, I ended up doing a joint PhD within Leiden and the University of Illinois at Urbana-Champaign (UIUC) and that was, as I said, an important part of my life because there were many people working on behaviour understanding and facial expression recognition. I was from the very beginning attracted to this field, and I started to work together with my colleagues there and so on. And ever since, I continued to work on this. Then, of course, I extended it to, let’s say, broader behaviour understanding and so on. In the end, after twenty years of working in this field, I would say that I have been involved in the most important conferences, the most important research directions that the community has done in the last twenty years. Then, slowly, slowly I started to build up my own group. I moved to Italy in 2009.
One of my main ways of interacting with and teaching my students is by challenging them to do impossible stuff. I remember that about ten years ago, I was telling my students that my ambition is to kind of animate me, let’s say, take a static image of myself and make me do, for example, the moonwalk of Michael Jackson. In the beginning, ten years ago, it was really looking like something from science fiction, but then, slowly, slowly, we really managed to do that. We started with, and that was one of the high impact projects we did a few years ago, a first-order motion model in which practically you have a static image, a driving video, an example would be Michael Jackson doing the moonwalk, and then the idea would be to take and animate this particular static image to do exactly the motion from the driving video. That was one of the first attempts towards synthetic media creation in a realistic way. We are not considering only avatars, but we are considering real images, real humans that could be animated and we worked in this direction.
I believe this is one of the main challenges at the moment and some of the main trends in the community, also because we are now having the right computational power and the right tools to provide this. The solutions that are at the moment are really, really incredible. I believe that in the last two to three years the community has done a big, big leap forward in this direction.
It is well known that you are a remarkable Machine Learning researcher with an extraordinary career. Can you share with us some of your current scientific activity at the University of Trento? Also, what can you tell us about the students you work with?
It’s no surprise that I got involved with Humans.ai simply because I’m working with my research group exactly in this area. As I said, I have been working on image and video creation, image animation and so on for the past few years, and I have many students working on this. One positive thing is that once the group has started to grow up, it was more like a recursive type of approach. You know, I have good students and because I have good students, we have good results. Since we have good results, there are good students applying to come into the group, and it keeps on going like that.
At the moment I have a group of about twenty people and what I like is that people have different seniority levels. So, at the top I have my former students, about twenty-five at the moment, who are now working in all the big labs, Google, Facebook, you name it, and they are also supervising the new students coming into the group. They can go and do internships and so on. Altogether, we are like a big family. So, my goal in research is to provide opportunities for young people to work together. I’m very happy to see that my group is like a big family. One other thing that I’m trying to have is a multiethnic, multinational group. In my group, I think I have, and I had people from everywhere. This is, let’s say, very unusual for Italy, where usually the research groups tend to be consisting only of Italians. I would say that my group is probably one of the most international in Italy.
I believe we are one of the most successful groups in computer vision and multimedia in Italy and in Europe. This is entirely due to the work of my students, and you know, they are the ones putting the effort and in an enthusiastic way providing all the methods and solutions. Of course, we are doing a lot of brainstorming and the fact that we are a big family helps a lot because people start to grow confident and to believe that they can do well. This is also reflected by the fact that people graduating from my group are literally hunted by the big multinationals to work in their labs.
What I’m trying to do also with Humans.ai is to bring some of my former students and collaborators to come and help and work with us. I believe that we can also build up this family here, together with some of my former students, with other people, so that we could really develop something very challenging and innovative.
You have one of the most coveted research jobs in the world. AI creators are, in a way, the astronauts of our generation. From your point of view, how do you relate to this responsibility of shaping the future of humanity or at least a part of the future of humanity?
Well, you know, we really don’t see it in that particular way in the sense that we do our job. To me, everything looks incremental, in the sense that, well, I’m building today on top of what I did yesterday, right? So, it’s true that looking from the outside, it looks like something very, very extraordinary, but for us, it’s more like regular business. What is probably missing in general for AI research is the fact that there is a gap between what we are doing at the moment and what the potential is, let’s say, towards applying what we do in society, community and so on. This is increasingly reducing at the moment because many academic researchers are also involved now with startups, companies and so on. It’s becoming more and more common to have joint positions between universities and research labs.
For me, joining Humans.ai was exactly in this direction, being able to put together some of the methods that we are developing as researchers, let’s say like toys if you want and to put them into real life and to try to see if we can address the needs of real customers. At the university, what we are doing is mostly creating research papers. This is good of course, and then we are going from deadline to deadline, but in some sense, and this I’m trying to tell my students is that we are somehow missing the connection/impact on society. This is one of the main reasons why I wanted to join Humans.ai. We try to make this a channel through which we can take the technology out of the academic sphere and direct it towards something useful to society.
How did you start working at Humans.ai? What made you join the project?
That’s a very good question. I believe I like the simple slogan that Humans.ai has adopted from the beginning, you know, the fact that we are several billion people on this planet and each of us could have his own unique contribution. I like this idea, and I believe in it. Of course, many of these people do not have an idea of how much their contribution is worth, but I think if we can just empower them to put together what they can, as little as they can and to put it together to achieve something useful in general for society, it will be very important and it is a strong message. I guess that’s the main reason why I like the idea. Also, I’m at a point in my career in which I don’t need anything else but to challenge myself by being involved in interesting and diverse projects like this one, including working with young people and young communities.
How would you describe the work environment at Humans.ai?
Well, I like it because it’s very informal. It’s also very good because the team as a whole is complementary and very diverse. We were talking before that one of the main ambitions of Humans.ai is to put NFT and AI together, and this is exactly reflected in the team. So, I believe the fact that we have people from different expertise put together, talking and trying to see how you can blend these two communities together is one of the main strengths of Human.ai. This is something that is happening, and I think this is the key to success.
What is your motto? What guides you through life?
Oh, it’s a good one. I haven’t thought about it, but I would say it is “Never stay quiet, always look for something else.”. People who know me would say that I’m a dynamic person. I would say I’m an extrovert, so I like to challenge myself, and I always like to find new, let’s say reach out to goals, and this is what I’m doing. I believe that I stay young because of that.
What else can you tell us about yourself, about the small pleasures of life? Do you have any hobbies?
Well, whenever I have the opportunity, I go skiing. One of the main reasons I wanted to move from Holland to Trento was exactly the fact that Trento is a mountain paradise. So, I go skiing as much as I can. I also go hiking and so on. I think that would be my main hobby. Well, that’s about it, I would say.