Interview with Manuela Veloso

Manuela Veloso

Manuela Veloso
Herbert A. Simon University Professor
Head, Machine Learning Department
School of Computer Science
(joint in CSD, courtesy in RI, MechE, and ECE)

Note: We’ve conducted an interview previously with Professor Veloso. Check it out here!

What is the biggest challenge you've overcome in your career?
That's hard. I don't know what is the biggest challenge I overcome in my career. So the career, the academic career is full of challenges, and so I don't know what's the biggest one, but a lot of challenges we face are related to the research we do.

And so it's about trying to really coming up with research ideas that are both interesting and novel enough. The challenges are related to the theses of the PhD students to be of great impact, so you want the student to be able to accomplish something novel and interesting within a limited amount of time.

So basically those are the challenges of our career in terms of advising students. And in terms of teaching, of course, the challenges are always to make sure that the students learn the material or learn the concept, but also become interested in the topics of, in my particular case, of artificial intelligence, machine learning, and robotics. So these are the challenges we face, I mean in terms of career, in terms of being successful. But I don't know what is the biggest.

What motivates you to work as hard as you do?
Very good question. So it's really self-motivation. I mean people want to do well and people want to – I mean at least I want to – advance the state of the art of AI, robotics, machine learning. So that motivates my research pursuits with my students. Also now that I am the Head of the Machine Learning, I am motivated to keep its reputation very high, to keep the best possible environment for the students, for the faculty, for the staff. So those goals motivate me to work hard.

The motivation is to push things forward and there are many things that we are pushing forward from the research all the way to the administrative aspect. So there is a wide range of things to work on, so I have to work hard. :-)

You hit on this during our February Student Faculty Lunch and we wanted to share some of your advice to a larger audience: could you hit on any advice that you have for students who are afraid of failure?
I think that people have to be able to accept that fear exists and not use the fear not to proceed forward. I think that everyone needs to try their best, and not necessarily be very dependent on the evaluation part, failure or success. We need to be a little bit more independent and more distant from the fact that it failed or succeeded, or other people like it or praise it or reject it. We need to gain our own internal evaluation and satisfaction.

The evaluation of what you cannot affect us as much, so we don't do to be evaluated in some sense; you do things because you are interested on the problems, because you want to do the best you can - that’s what drives you.There is only failure if you are afraid of not being – even if you are afraid of people saying you failed, but if by any chance you are afraid of not being able to solve that problem, okay, the best thing you can do is try to solve it. And if you don't solve it, well, maybe you'll solve the next one.

It's a little bit more about compensating for the fear with doing. And then if you don't do it, well, too bad. But the fear is not what the focus should be on; the focus should be on trying and doing. It takes time to get to control the fear, but we get better at it, and it’s worth trying to focus on solving the problem in our best possible way, independently of the external evaluation.

Can you talk a little bit about what you think the future holds for machine learning?
This is something that is very important to understand. Machine learning is indeed a discipline that will be of increasing relevance throughout our lives. Because as we increasingly collect so much data in the world - from weather data to our shopping data to our health data to the traffic data - we will increasingly need automated algorithms to analyze such data. We humans cannot process all that digital data. Machine learning is this process of actually automating the data analysis process. And then in addition, machine learning and AI also can help with decision-making based on the data. People will be able to use machine learning and AI to support their decisions. Classification, prediction, optimization, decision-making are data driven and in the realm of the space of machine learning.

So machine learning comes inevitably or comes attached to the fact that we are digitalizing so much data, and therefore so much data cannot be processed in any other way - by anyone manually, and therefore this is not the problem of just algorithms, this is not the problem of computer architecture, it's not the problem of just signal processing; it's the actual understanding of that data, the extraction of patterns of the data, which is in fact completely part of what the discipline of machine learning is all about. In addition to the decision-making. So it's improving AI systems, improving automated systems with experience. So like recommended systems, they do better over time. So we cannot stop doing machine learning unless we would stop collecting data, which isn't happening. So that's why machine learning is so important, because of these enormous amounts of data. So then who is going to understand all that data, if not machine learning and AI algorithms?

What do you like to do in your free time?
The thing I like most to do in my free time is to be with my family, with my husband, my parents, my children, my grandchildren, my brothers and sisters, my nephews and my nieces. When and if I have free time, I used it all to be with my family, as much as possible.

I like cooking, I like watching movies from time to time. And I also like to exercise - now just walking and swimming in the summer.