U of T Statistical Sciences: Can you tell us about how you got interested in statistics and your academic journey to getting a PhD?
Cédric Beaulac: I always knew I wanted to study mathematics. When I was younger, I was concerned about getting a job if I studied theoretical sciences, so statistics felt like a better option, since I knew that it would be easy to find a job in the end.
How did you decide on doing a PhD?
After I completed my master’s degree, I wrote a thesis. The process of writing my thesis really confirmed that I enjoy doing research. I had a ton of fun and just wanted to continue that with a PhD.
I was also curious to test my personal limits, so I thought, "If I can push forward, why not?”
Can you tell us a bit more about your PhD thesis? What was it about?
I knew I wanted to do something related to machine learning. I've always dreamt of working in Artificial Intelligence. Artificial intelligence is a very large field, so I tried to explore the machine learning side of it, which is mostly the computer science approach to data analysis. It was a nice way to get to know that field and to learn about AI these days.
My thesis is the exploration of machine learning through the lens of a statistician, which includes experiments using machine learning algorithms in various statistical applications, such as medical science, higher education, and image analysis.
Fundamentally speaking, I think that statistics and machine learning are doing the same thing. They are both trying to solve data analysis problems. In most cases you can boil down a problem to supervised learning problems. We have a bunch of inputs and we're trying to predict an output. We are looking at the function that relates those inputs to an output, and we have to train or fit that function in in some way.
So, the question I was asking in my thesis was: can we use machine learning algorithms to solve various statistical problems?
As someone who has a broad view on machine learning, can you tell us something that the public doesn't really understand about machine learning? Or a fact that is not well known?
I'd say machine learning is not as complicated as you may think. Also, I think that people sometimes think it’s a magical formula that will instantly solve all of our modern problems. I don't think machine learning is an instant solution that will solve everything.
What does the day-to-day life of a PhD student look like?
Except for the first year, which is similar to an undergraduate lifestyle where you have courses and homework, you kind of become your own company as a PhD student.
For me that looked like this: waking up, drinking a coffee while taking a look at my emails, maintaining my website, attending meetings. Once I got all the paperwork and extra-curricular stuff out of the way, I could focus on research. A couple of days a week I read articles or books related to my research and tried to stay in touch with the field. I did some coding, attended some talks.
The best part about being a PhD is that you can make your own schedule and work on your own projects.
What advice do you have for students who are either interested in statistics and machine learning or students who are interested in pursuing a PhD in this field?
This might be controversial, but I'd say if you can do a master's degree with a thesis, you should. I’m so glad I did that, and it helped me greatly in my PhD program.
Generally speaking, doing a PhD is definitely not the fastest way to get rich, but it’s a really rewarding personal challenge, so make sure that’s something that motivates you.
Can you talk a bit about your postdoc that you're starting this year and what you're looking forward to?
I actually have a postdoc and I have a teaching position or a tenure-track position at a University in Quebec already lined up. The postdoc is in British Columbia. I'll be working under the supervision of Farouk Nathoo and Faisal Beg. Farouk is at the University of Victoria right now and Faisal is a professor at Simon Fraser University
They do genetic imaging. What we try to do is predict or understand the causal relationship between a set of genes and an illness.
We work with MRI scans of the brain and right now we're looking at Alzheimer's. We know Alzheimer's is a genetic disease. So, the question is can we identify the gene that causes Alzheimer's?
We try to match the genetic codes to the patient status, but in this particular case we have actual brain scan images and we're trying to map or equate those genetic markers to an actual brain scan.
This is really aligned with my new passion for image analysis, so we're going to try to solve the mystery of Alzheimer's.
Based on your own experience, what advice do you have for students on finding their research interests?
It is very important that you really enjoy your research. If you end up working on a project that you don’t enjoy, you will struggle to stay motivated during those long hours of reading and writing.
The good news is that as a PhD student you have plenty of time in your first year to explore: go to conferences, go to talks, and meet professors that are willing to support your research. There's no need to rush it. You have plenty of time to find your own path and what you’re good at.
.jpeg?auto=webp)
.jpeg?auto=webp)
/1x.jpeg?auto=webp)
/1x.jpeg?auto=webp)
/1x.jpeg?auto=webp)
/1x.jpeg?auto=webp)
/1x.jpeg?auto=webp)
/1x.jpeg?auto=webp)