Improving Resnet-9 Generalization Trained on Small Datasets

TL;DR

This paper introduces a novel training approach for ResNet-9 that significantly improves generalization on small datasets, achieving high accuracy rapidly and winning a hardware-aware training competition.

Contribution

The paper presents a combination of techniques including sharpness aware optimization and meta-learning to enhance ResNet-9's performance on limited data within strict time constraints.

Findings

Achieved 88% accuracy on small CIFAR-10 subset

Training time under 10 minutes

Won ICLR competition on hardware-aware efficient training

Abstract

This paper presents our proposed approach that won the first prize at the ICLR competition on Hardware Aware Efficient Training. The challenge is to achieve the highest possible accuracy in an image classification task in less than 10 minutes. The training is done on a small dataset of 5000 images picked randomly from CIFAR-10 dataset. The evaluation is performed by the competition organizers on a secret dataset with 1000 images of the same size. Our approach includes applying a series of technique for improving the generalization of ResNet-9 including: sharpness aware optimization, label smoothing, gradient centralization, input patch whitening as well as metalearning based training. Our experiments show that the ResNet-9 can achieve the accuracy of 88% while trained only on a 10% subset of CIFAR-10 dataset in less than 10 minuets