Toki Tahmid Inan

EvoBatch: A Mini-Batch, Compute-Constrained Evolutionary Algorithm Reduces the Generalization Gap in Deep Learning

EvoBatch is introduced, a novel Hybrid Evolutionary Algorithm that re-frames deep network optimization as an explicitly multi-objective problem, and establishes that compute-constrained, multi-objective evolutionary optimization offers both an effective and an efficient alternative to single-objective gradient methods...