Near-Optimal Cryptographic Hardness of Agnostically Learning Halfspaces and ReLU Regression under Gaussian Marginals.

Ilias Diakonikolas, Daniel Kane, Lisheng Ren

Browse the full ICML paper archive.