This paper explores critical learning periods in deep linear network models and shows that these periods depend on the depth of the model and structure of the data distribution. The study also examines the impact of pre-training on transfer performance in multi-task learning.

https://arxiv.org/abs//2308.12221

YouTube: https://www.youtube.com/@ArxivPapers

PODCASTS:

Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016

Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers