The Statistics Department is featuring four statisticians over several months as part of our series celebrating the inaugural International Day of Women in Statistics and Data Science that occurred in October (October 11th, 2022). Introduced by the Caucus for Women in Statistics, the Portuguese Statistical Society, and the American Statistical Association, the day celebrates the research contributions of women statisticians and data scientists around the world.
While our last interview featured Professor Xihong Lin, the following excerpted and revised interview is with Assistant Professor Morgane Austern. During our conversation with Prof. Austern, we learned more about her research interests in machine learning and dependent structured data, the conversations required to retain women in the field, and we even heard about her long-term fascination with whales!
1. Please introduce yourself (your position at Harvard and a summary of your current areas of research).
Austern: My area of research is applied probability and machine learning theory, which includes understanding the fundamental principles underlying machine learning and statistical learning. More specifically, I am interested in how structured data [structured data exists in a format, as opposed to “unstructured data”] and the dependence of data affect the performance of algorithms.
To explain structured data and dependent data a little more, let’s imagine that you want to understand climate change by studying weather data. In this case, the data you collect would be called stochastic dependent data because, for example, the weather of today will influence the weather of tomorrow. This is in contrast with the traditional assumption that data comes in the form of a sequence of independent observation; for instance, if you were examining the average height for an American, you would sample Americans’ height at random and each data point would be independent. I research how models can be trained on dependent and structured data to build good machine learning algorithms. I want to know how these machine learning algorithms behave, what makes them work, and whether we can give theoretical guarantees for them.
2. What inspired you to go into this field and what motivates you in your everyday work? When you look back on your path as a researcher, was there an experience and/or mentor that stood out?
Austern: There wasn’t a special moment of inspiration that drove me to be a researcher in statistics - it was simply that I liked math. I liked thinking. I liked asking questions. My mom enjoys sharing anecdotes about my interest in math as a kid, even before I started formal schooling. She would tell you that my curiosity was prompted by her calculator, which had a square root symbol that I would ask questions about (this is probably not the whole truth!).
While it’s difficult to pinpoint a single mentor or moment that influenced my career path, I had a conversation that helped me to consider a Ph.D. in statistics. As a young, naïve student, I was planning on doing something completely different from a Ph.D. (something that I probably would have hated!). When I asked for a letter of recommendation from a professor who had taught me advanced probability, she said, “You know what, I think you are very good at probability and that you shouldn’t do what you want to do. I will write you a letter of recommendation on the one condition that you also apply to a Ph.D. program.” At the time, I thought, why? No one in my family except my grandfather had earned a Ph.D. After thinking about our conversation more, I followed her advice by applying to one Ph.D. program, which accepted me. Once I began graduate school, I started pursuing interesting research questions and realized that I preferred this type of work in an academic setting.
3. Describe a recent project that has been exciting to work on. Why is this an important project?
Austern: As I mentioned, one research area of interest to me is machine learning and dependent structured data. Traditionally, to give guarantees for statistical learning and machine learning, you would set up a sequence of observation that results in gathering independent data. To analyze this independent data, you would rely on numerous off the shelf probability theorems, which hold under very general conditions and are essential for quantifying uncertainty in statistical learning. I’ve been working on an exciting, multi-year project with two students to examine how we can generalize the most basic, useful toolkit of probability theorems in an independent setting to a large class of dependent, structured objects. Normally, when studying a large class of objects, like a network or graph, you would need to generate a series of papers generalizing a toolkit for a specific structure. However, I am exploring how to build a bigger framework in which a new theorem could account for most structures. While other scholars are working in this area, I approach the research from a different angle by combining different mathematical tools, such as ergodic theory and functional analysis.
4. What steps do you think are effective for recruiting and retaining women in your field?
Austern: First, I would like to emphasize that it’s great to be a woman in statistics and there has never been a time that’s been better. Also, I absolutely love where I am – this Department has a very supportive atmosphere. However, to continue to attract and retain women statisticians, we still need to address a lot of issues in the broader scientific community. In the larger profession, problems arise such as sexual harassment at conferences, including inappropriate remarks; I don’t know any faculty member or student who is unfamiliar with this reality, unfortunately. Through our professional organizations, we need to have more conversations about how to prevent and handle inappropriate behavior. At the same time, I’m hopeful because there is a lot more awareness about sexual harassment and discrimination today than when I was in a Ph.D. program.
Another crucial conversation that should be more normalized is how to develop a positive work-life balance, an important factor for retaining women and men. I feel very supported here in the Department and in the University with the conversations that I’ve had on this topic, but this kind of support is less uniform in the broader scientific community. To ensure that young researchers feel welcomed into the field, we need to talk about this more openly and recognize people’s full identities; we are mathematicians, but we are also human beings and care about many things. My hope would be that any Ph.D. student would feel comfortable asking their advisor, How does it work if I want to have kids some day and be a professor?, but I don’t think we are there yet.
My graduate program experience helped me to grapple with some of these issues in a support group for women in my program. After seeing the benefits of this group, I recently introduced a Women in Statistics Group in our department. Through this group, I aim to offer a positive space for women-identifying individuals in the department to share questions and concerns and build community, which in turn can help with retention and overall satisfaction with their experience.
5. What advice would you give to young women (in grade school, college, and beyond) who are interested in pursuing research in your field?
Austern: There are a lot of cool opportunities and careers and interesting questions that you can pursue with math – so you should do it, if you love it! I would give the same advice to young women and men to pursue math if they are interested in it, and, if young women are told that they shouldn’t go into math (because of gender expectations), they should ignore this sentiment. Math is for everyone. If young students want to become researchers in statistics or other math-related fields, they should also know that this type of work requires you to be comfortable with not knowing the answers and with persevering through some tough challenges.
To younger students, I would also say that you can be a great statistician or mathematician, and it can be the right career for you even if it isn’t your one passion. I’m not passionate about just one area, instead I care about many things – I could bug you about history or details about whales for hours but that doesn’t mean that I shouldn’t do math. As a kid, I was always curious about many topics, so while I loved math and read about math, I also read about history and biology. Passion should be encouraged in young students but shouldn’t be the only thing to aspire to when considering a career in math or statistics.
Stats: Prof. Austern’s insights into how to encourage young scholars to pursue (and stay in) careers in math and academia reveal that, while there is still much work to be done, there are concrete improvements that can be made today. We thank Prof. Austern for sharing her time with us.
Please stay tuned for the next interview in our series with Assistant Professor Tracy Ke.