The PLAI group research generally focuses on machine learning and probabilistic programming applications. Our aim is to develop foundational knowledge and tools in this area, to support existing interest in different applications.

Probabilistic programming languages (PPL) are on the cusp of becoming practically useful for expressing and solving a wide-range of model-based statistical reasoning problems. Continuing PPL research and development will make it possible for the AI community to rapidly develop key new models for perception, reasoning, and action selection that go beyond what current deep learning systems are capable of.

Through “inference compilation”, we can bridge the gap between probabilistic programming and deep learning, leveraging deep learning techniques to dramatically speed repeated inference in richly structured models denoted in the form of probabilistic programs. 

With development and maturation of such languages, it will become possible to more rapidly and efficiently explore an exciting spectrum of applied AI research projects using amortized inference, e.g. simultaneous agent-based inference and planning using, for instance, partially specified video game-like engines for rendering and simulation. In other words, significant new progress beyond deep learning and towards a new tool-chain for AI.

To explore some of our current projects, us the links on the left of the screen.

Supported By