English

EvoGrad: Efficient Gradient-Based Meta-Learning and Hyperparameter Optimization

Machine Learning 2021-10-28 v2 Neural and Evolutionary Computing Machine Learning

Abstract

Gradient-based meta-learning and hyperparameter optimization have seen significant progress recently, enabling practical end-to-end training of neural networks together with many hyperparameters. Nevertheless, existing approaches are relatively expensive as they need to compute second-order derivatives and store a longer computational graph. This cost prevents scaling them to larger network architectures. We present EvoGrad, a new approach to meta-learning that draws upon evolutionary techniques to more efficiently compute hypergradients. EvoGrad estimates hypergradient with respect to hyperparameters without calculating second-order gradients, or storing a longer computational graph, leading to significant improvements in efficiency. We evaluate EvoGrad on three substantial recent meta-learning applications, namely cross-domain few-shot learning with feature-wise transformations, noisy label learning with Meta-Weight-Net and low-resource cross-lingual learning with meta representation transformation. The results show that EvoGrad significantly improves efficiency and enables scaling meta-learning to bigger architectures such as from ResNet10 to ResNet34.

Keywords

Cite

@article{arxiv.2106.10575,
  title  = {EvoGrad: Efficient Gradient-Based Meta-Learning and Hyperparameter Optimization},
  author = {Ondrej Bohdal and Yongxin Yang and Timothy Hospedales},
  journal= {arXiv preprint arXiv:2106.10575},
  year   = {2021}
}

Comments

Accepted at NeurIPS 2021