Feasible Learning (FL) optimizes model performance on individual samples, using a primal-dual approach to enhance training dynamics, showing improved tail behavior over Empirical Risk Minimization with minimal average performance impact.

https://arxiv.org/abs//2501.14912

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

TikTok: https://www.tiktok.com/@arxiv_papers

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

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