Second-order Optimization Method for Large Mini-batch: Training ResNet-50 on ImageNet in 35 Epochs
share
Large-scale distributed training of deep neural networks suffer from the generalization gap caused by the increase in the effective mini-batch size. Previous approaches try to solve this problem by varying the learning rate and batch size over epochs and layers, or some ad hoc modification of the batch normalization. We propose an alternative approach using a second-order optimization method that shows similar generalization capability to first-order methods, but converges faster and can handle larger mini-batches. To test our method on a benchmark where highly optimized first-order methods are available as references, we train ResNet-50 on ImageNet. We converged to 75 validation accuracy in 35 epochs for mini-batch sizes under 16,384, and achieved 75
READ FULL TEXT