Events

 
<< May 2021 >>
 
 >>
 >>
 >>
 >>
 >>
 >>
Sun Mon Tue Wed Thu Fri Sat
   1 
 2   3   4   5   6   7   8 
 9   10   11   12   13   14   15 
 16   17   18   19   20   21   22 
 23   24   25   26   27   28   29 
 30   31   
 
Add an Event Edit This Event
<< Spring 2021 Summer 2021 Fall 2021 >>
Subscribe your calendar or receive email announcements of events
Physics ∩ ML Seminar
Are wider nets better given the same number of parameters?
Date: Wednesday, May 19th
Time: 11:00 am - 12:15 pm
Place: Online Seminar: Please sign up for our mailing list at www.physicsmeetsml.org for zoom link
Speaker: Anna Golubeva, Perimeter Institute
Abstract: Empirical studies demonstrate that the performance of neural networks improves with increasing number of parameters. In most of these studies, the number of parameters is increased by increasing the network width. This begs the question: Is the observed improvement due to the larger number of parameters, or is it due to the larger width itself? We compare different ways of increasing model width while keeping the number of parameters constant. We show that for models initialized with a random, static sparsity pattern in the weight tensors, network width is the determining factor for good performance, while the number of weights is secondary, as long as the model achieves high training accuracy. As a step towards understanding this effect, we analyze these models in the framework of Gaussian Process kernels. We find that the distance between the sparse finite-width model kernel and the infinite-width kernel at initialization is indicative of model performance.
Host: Gary Shiu
Add this event to your calendar