Tamara Broderick first set foot on the MIT campus as a high school student, as a participant in the inaugural conference Women's Technology Program. The month-long summer college experience gives young women a hands-on introduction to engineering and computer science.
How likely is it that she will return to MIT years later, this time as a faculty member?
This is a question that Broderick could likely answer quantitatively using Bayesian inference, a statistical approach to probability that attempts to quantify uncertainty by continually updating one's hypotheses as new data is obtained.
In her MIT lab, the new associate professor in the Department of Electrical Engineering and Computer Science (EECS) uses Bayesian inference to quantify uncertainty and measure the robustness of data analysis techniques.
“I always wanted to understand not only 'What do we know about data analysis,' but also 'How much do we know it?' “, says Broderick, who is also a member of the Information and Decision Systems Laboratory and the Institute for Data, Systems and Society. “The reality is that we live in a noisy world and we can't always get exactly the data we want. How can we learn from data while recognizing that there are limitations and how to manage them appropriately? »
Broadly speaking, its goal is to help people understand the limitations of the statistical tools available to them and, sometimes, to work with them to create better tools suited to a particular situation.
For example, his group recently collaborated with oceanographers to develop a machine learning model capable of rendering more accurate predictions about ocean currents. In another project, she and others worked with degenerative disease specialists on a tool that helps people with severe motor disabilities use a computer's graphical user interface by operating a single switch.
A common thread woven through his work is an emphasis on collaboration.
“Working in data analytics, you get to spend time in everyone's backyard, so to speak. You really can’t get bored because you can always learn about another area and think about how we can apply machine learning there,” she says.
Hanging out in many college “backyards” is particularly appealing to Broderick, who struggled even from a young age to hone his interests.
A mathematical mindset
Growing up in a suburb of Cleveland, Ohio, Broderick has been interested in mathematics for as long as she can remember. She remembers being fascinated by the idea of what would happen if you kept adding a number to itself, starting with 1+1=2 and then 2+2=4.
“I was maybe 5, so I didn't know what 'powers of two' were or anything like that. I was really passionate about mathematics,” she says.
Her father recognized her interest in the subject and enrolled her in a Johns Hopkins program called the Center for Talented Youth, which gave Broderick the opportunity to take three-week summer courses on a range of subjects, from astronomy to number theory and computer science.
Later, in high school, she conducted research in astrophysics with a postdoctoral fellow at Case Western University. In the summer of 2002, she spent four weeks at MIT as a member of the inaugural class of the Women's Technology Program.
She particularly appreciated the freedom the program offered and its emphasis on using intuition and ingenuity to achieve high-level goals. For example, the cohort was tasked with building a device with LEGOs that they could use to biopsy a grape suspended in Jell-O.
The program showed him how much creativity is involved in engineering and computer science and sparked his interest in pursuing an academic career.
“But when I got to college at Princeton, I couldn't decide: math, physics, computer science, they all seemed really cool to me. I wanted to do everything,” she says.
She decided to pursue an undergraduate degree in mathematics, but took every physics and computer science course she could fit into her schedule.
Dig into data analysis
After receiving a Marshall Scholarship, Broderick spent two years at the University of Cambridge in the United Kingdom, earning a Master of Advanced Study in Mathematics and a Master of Philosophy in Physics.
In the UK, she completed a number of statistics and data analysis courses, including her first course on Bayesian data analysis in machine learning.
It was a transformative experience, she recalls.
“During my time in the UK, I realized that I really enjoyed solving real-world problems that matter to people, and Bayesian inference was being used in some of the most important problems,” she says.
Returning to the United States, Broderick traveled to the University of California, Berkeley, where she joined the laboratory of Professor Michael I. Jordan as a graduate student. She earned a doctorate in statistics with a specialization in Bayesian data analysis.
She decided to pursue an academic career and was attracted to MIT by the collaborative nature of the EECS department and the passion and kindness of her future colleagues.
His first impressions were positive, and Broderick says he found a community at MIT that helps him get creative and explore difficult, impactful problems with wide-ranging applications.
“I have been fortunate to work in my lab with a truly wonderful group of students and postdocs: bright, hard-working people whose hearts are in the right place,” she says.
One of his team's recent projects involves a collaboration with an economist who studies the use of microcredit, or the lending of small amounts of money at very low interest rates, in poor areas.
The goal of microcredit programs is to lift people out of poverty. Economists conduct randomized controlled trials in villages in a region that receive or do not receive microcredit. They wish to generalize the results of the study, predicting the expected result if microcredit is applied to other villages outside their study.
But Broderick and colleagues found that the results of some microcredit studies can be very fragile. Removing one or a few data points from the dataset can completely change the results. One problem is that researchers often use empirical averages, in which a few very high or low data points can skew the results.
Using machine learning, she and her collaborators developed a method to determine how many data points need to be removed to change the study's substantive conclusion. Thanks to his tool, a scientist can see how fragile the results are.
“Sometimes removing a very small fraction of data can change the main results of a data analysis, and we may then be concerned about the extent to which these conclusions generalize to new scenarios. Are there ways to point this out to people? This is what we are aiming for with this work,” she explains.
At the same time, she continues to collaborate with researchers in various fields, such as genetics, to understand the advantages and disadvantages of different machine learning techniques and other data analysis tools.
Good road
Exploration is what motivates Broderick as a researcher, and it also fuels one of her passions outside of the lab. She and her husband enjoy collecting the plots they earn from hiking every trail in a park or trail system.
“I think my hobby really combines my interests in the outdoors and spreadsheets,” she says. “With these hiking areas, you have to explore everything, and then you discover areas you wouldn't normally see. It's adventurous, that way.
They discovered some incredible hikes they would never have heard of, but also embarked on several “completely disastrous hikes,” she says. But every hike, whether a hidden gem or an overgrown mess, offers its own rewards.
And just like in her research, curiosity, open-mindedness and passion for problem-solving never led her astray.