Reproduce “Implicit Bias of SGD in Multivariate ReLU Networks: Effective Width Collapse” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shuang Liang, Tom Jacobs, Guido Montúfar
The authors study feature selection in two-layer ReLU regression networks. In the mean-field limit, qualifying predictors are piecewise affine, and, under an additional 2-Wasserstein convergence condition, input weights and biases align along finitely many directions. Kink hyperplanes number at most 2P(X)-1, where P(X) counts strictly separable partitions of distinct training inputs; this structural bound is not a generalization guarantee.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.