Assistant Professor Frederic Sala joins UW-Madison Computer Sciences to teach Machine Learning and Data Analytics

Frederic SalaFrederic Sala joins the UW-Madison Computer Sciences Department from Stanford University where he was a postdoctoral scholar. He completed his PhD in electrical engineering at University of California-Los Angeles, where he studied machine learning, information theory, and data science and their intersection. His goal through his research here at UW is to help data-driven systems – such as using Siri and loan applications – be “reasonable, functional, and beneficial.” Sala returns to the Midwest after a decade in California and is looking forward to the four seasons: skiing, biking, and enjoying the lakes.

Hometown: I mostly grew up in the Midwest, especially in Southeast Michigan.

Educational/professional background: I did my PhD at UCLA in electrical engineering; afterwards, I was a postdoctoral scholar in Stanford’s Computer Science Department.

How did you get into your field of research? When I was an undergraduate at the University of Michigan, each time I went into the electrical engineering building, I passed by a statue of Claude Shannon. The first fifty times I didn’t pay any attention to it, but for some reason at one point I became curious and looked up his biography. Shannon is the father of information theory, the field that studies how data is communicated and stored, and how it can be compressed and protected from noise. I got excited by this field, which has a bunch of practical accomplishments that made the modern world of communications possible, but also has a really beautiful underlying mathematical basis. Eventually I worked on information theory in grad school. As I was working on a way to prevent noise from affecting any aspect of data being stored in memories, I came to see that a lot of this data is noisy to start with, and that might actually be OK to ignore some types of noise. Machine learning, in particular, deals with noisy data all the time. This roundabout process—should we protect against noise, but only sometimes, and how do we know when it matters, got me started in machine learning.

Could you please describe your area of focus? I work on understanding the fundamentals of systems that are driven by data, such as those in machine learning and data analytics. I study the underpinnings of these systems. A lot of my focus is on understanding the connections and interplay between different aspects of these systems—the nature of the data that goes in, the way components fit together, and the requirements for the output. My goal is to try to understand the limits of such systems, and to make those insights practically useful.

What main issue do you address or problem do you seek to solve in your work? Machine learning and data analytics are now ubiquitous; they form part of so many different systems that impact our lives in all sorts of ways. One of the challenges is that we often don’t have a concrete way to understand when such systems should and shouldn’t work. Often, the basic explanation for why something works doesn’t quite apply to the setting we face in real-world usage. For example, what if we don’t have tons of labeled data, but do have some side information that we can express in a certain way? How reliable will such systems be? When can we say confidently that one approach won’t work, or another will? These are the types of problems I enjoy studying.

What’s one thing you hope students who take a class with you will come away with? Understanding the intuition behind why things work. Machine learning involves a lot of different areas—probability, linear algebra, statistics, optimization, information theory, programming languages, etc. It’s easy to lose track of the big picture when there’s some many parts involved in making the machine run. It’s important to keep track of how each part fits together and the intuitive reason for why they’ll all work.

What attracted you to UW-Madison? It’s a wonderful university, department, and faculty. Everyone here is brilliant and kind, and I was excited to join. In addition, after a decade in California, I was ready to enjoy a beautiful four-season climate.

What was your first visit to campus like? My first visit was when I moved to Madison! Due to Covid-19, I wasn’t able to take the usual visits. I moved here during the middle of fall, with the foliage at its best. It was wonderful just walking around on the lakeshore.

What are you looking forward to doing or experiencing in Madison? I’m excited to get back into winter sports, especially skiing. When the weather warms up, I can’t wait to sit at the terrace, ride my bike around the state, and have picnics with my partner.

Do you feel your work relates in any way to the Wisconsin Idea? If so, please describe how. We’re all at least a little bit dependent on data-driven systems for much of our daily lives—whether querying a search engine, asking Siri a question, or applying for a loan. Everyone deserves to have confidence that these systems are reasonable, functional, and beneficial. My goal is to help produce this outcome.

Hobbies/other interests: I like running, hiking, cycling, and reading. Dilettante in political and social sciences.