PoP Seminar Talk

Automatic Integration and Differentiation of Probabilistic Programs

Alexander Lew, Assistant Professor, Yale University
Monday, 16 March, 2026; 11:00am
NSH 3305
Host: Feras Saad

Abstract

By automating the error-prone math behind deep learning, systems such as TensorFlow and PyTorch have supercharged machine learning research, empowering hundreds of thousands of practitioners to rapidly explore the design space of neural network architectures and training algorithms. This talk will show how new programming language techniques—particularly generalizations of automatic differentiation and of knowledge compilation—make it possible to generalize and extend such systems to support probabilistic models. Our tools can automate the computation of expected values, probability densities, and their gradients, as well as help users derive fast, low-variance, unbiased estimators of these quantities when they are too expensive to compute exactly, enabling orders-of-magnitude speedups in downstream optimization and inference problems.

Bio

Alex Lew is an Assistant Professor of Computer Science at Yale. His research aims to automate and scale up principled probabilistic reasoning, drawing on techniques from programming languages, machine learning, Bayesian statistics, and cognitive science. A key focus is the theory, design, and implementation of probabilistic and differentiable programming languages, which extend traditional programming languages with constructs for optimization and inference over models defined as programs. Alex’s work has been recognized with Distinguished Paper awards at POPL and LICS, an Outstanding Paper award at the Conference on Language Modeling (COLM), and a Facebook Research Award.