Pure Noise to the Rescue of Insufficient Data: Improving Imbalanced Classification by Training on Random Noise Images.

Shiran Zada, Itay Benou, Michal Irani

Browse the full ICML paper archive.