Events

Colloquium

The Relative Value of Prediction

Speaker: Juan Carlos Perdomo Silva (NYU) Incoming Assistant Professor of Computer Science and Data Science New York University Wednesday, November 19, 2025 11:30AM - 1:00PM Lunch will be at 11:30am in room 1307 Talk will be at 12:00pm in room 1327 Location: 17 Hillhouse Ave, 3rd floor, , , and via Webcast: https://yale.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=1f44480f-7cd7-4684-9c23-b391015162a4

Talk Summary: Predictive algorithms are increasingly used to guide the allocation of scarce resources—from deciding which students receive tutoring in Wisconsin to targeting cash transfers in the developing world. Yet, predictions in these contexts are only a means to an end: they help planners make better decisions with the ultimate goal of improving social welfare (increasing graduation rates, reducing poverty, etc.). And viewed from this perspective, it’s clear that they are only a small piece of the puzzle. There are many other ways to improve welfare beyond increasing the accuracy of our predictive systems.

Given this broad design space of things we could do, the goal of this talk is to understand: when are investments in prediction truly “worth it”? I will discuss a new line of work that provides formal foundations for this question. Based on joint work with Christoph Kern and Unai Fischer-Abaigar.

Speaker bio: Juan Carlos Perdomo is a postdoctoral fellow at MIT and an incoming assistant professor of computer science and data science at New York University. His research focuses on the foundations of machine learning algorithms that make predictions or decisions about people. He earned his Ph.D. in EECS from UC Berkeley, where he was co-advised by Peter Bartlett and Moritz Hardt, and his B.A. in CS and Math from Harvard. He was also previously a postdoc at Harvard hosted by Cynthia Dwork.

Add To: Google Calendar | Outlook | iCal File

• Colloquium