Yoshua Bengio- System 2 GFlowNets- Induction Biases Towards Machines that Understand and Think

Explore a 30-minute conference talk by Yoshua Bengio, Professor of Computer Science at Université de Montréal and Founder & Scientific Director of the Mila AI Institute, on System 2 GFlowNets and induction biases towards machines that understand and think. Delve into topics such as deep learning, sparse dependencies, global workspace theory, modularity, and attention. Gain insights into upcoming papers, recent research, and the concept of discretization in AI. Discover the connections between classical AI and GFlowNets, and understand the high-level picture and training processes involved. This talk, presented at the Santa Fe Institute, offers a comprehensive exploration of cutting-edge AI concepts and their implications for developing machines with enhanced understanding and reasoning capabilities.

Intro
Summary
Distribution
System 1 System 2
Deep Learning
Sparse Dependencies
Global Workspace Theory
Modularity and Attention
Upcoming papers
Recent paper
Discretization
Classical AI
GFlowNets
High Level Picture
Training