Towards a General Theory of Infinite-Width Limits of Neural Classifiers
If you attend #icml2020, don't miss @GolikovEugene presentation at Deep Learning - Theory Section. Come to see paper about grasp an infinity width nets by general framework that allows to unify well known MF and NTK limits. Discussions will be held on Wednesday at 10 p.m. and Thursday at 9 a.m. Schedule: https://icml.cc/virtual/2020/poster/6845# Full paper: https://arxiv.org/abs/2003.05884