How to Train Your Wide Neural Network Without Backprop: An Input-Weight Alignment Perspective.

Akhilan Boopathy, Ila Fiete

Browse the full ICML paper archive.