Inductive biases for deep learning over sequential data: from connectivity to memory addressing

Speaker: 
Guillaume Lajoie, MILA/Universite De Montreal
Event time: 
Monday, January 18, 2021 - 2:30pm
Location: 
Zoom Meeting ID: 97670014308 See map
Event description: 

ABSTRACT:  In neural networks, a key hurdle for efficient learning involving sequential data is ensuring good signal propagation over long timescales, while simultaneously allowing systems to be expressive enough to implement complex computations. The brain has evolved to tackle this problem on different scales, and deriving architectural inductive biases based on these strategies can help design better AI systems.

In this talk, I will present two examples of such inductive biases for recurrent neural networks with and without self-attention. In the first, we propose a novel connectivity structure based on « hidden feed forward » features, using an efficient parametrization of connectivity matrices based on the Schur decomposition. In the second, we present a formal analysis of how self-attention affects gradient propagation in recurrent networks, and prove that it mitigates the problem of vanishing gradients when trying to capture long-term dependencies.

contact tatianna.curtis@yale.edu for password.

Event Type: 
Applied Mathematics