BN-invariant Sharpness Regularizes the Training Model to Better Generalization.

Mingyang Yi, Huishuai Zhang, Wei Chen, Zhi-Ming Ma, Tie-Yan Liu

Browse the full IJCAI paper archive.