Our group carries out research in machine learning oriented toward statistical principles and methodology.
What is machine learning? It is closely aligned with traditional statistics, but has a distinct focus on computation, prediction, and representation of data. Machine learning is also associated with complex problems like machine translation and image understanding. These can sometimes be broken down into concrete statistical estimation and inference problems that are studied in isolation.
Machine learning is increasingly conflated with artificial intelligence, as both are becoming household terms through mainstream media. One distinction to make is that machine learning is focused on prediction and inference, while an AI system will typically include a decision making component, and exhibit a behavior through the collective decisions that are made.
Rapid progress in machine learning, and ultimately AI, has come from years of hard work on formulating simplified learning frameworks and optimization problems in concrete mathematical terms. Progress has also come from a competitive spirit on grand challenge problems, coupled with broad collaboration and code sharing. The remarkable results achieved by deep neural networks can partly be seen as resulting from orders of magnitude more data and computational resources than were previously available, combined with powerful code bases that allow rapid experimentation. But this is just one part of a field that is full of creative ideas, and a deep understanding of machine learning systems is still emerging.
In his book Conscilience, Edward O. Wilson paints an "atoms to brains" portrait of the unified pursuit of knowledge across the sciences and humanities. There is a kind of conscilience in the way that data science, including modern statistics and machine learning, is becoming a primary means of acquiring this knowledge. All of this makes for an exciting research frontier that cuts across many different areas. It has something for everyone, from fundamental problems of pure mathematics to questions of public policy and societal impact.
The people in our research group are passionate about contributing to the development of the field. Related research is being done in several other parts of Yale. All are welcome to join our group meetings to learn more.