ES-MAML: Simple Hessian-Free Meta Learning
From MaRDI portal
Publication:6326439
arXiv1910.01215MaRDI QIDQ6326439FDOQ6326439
Authors: Xingyou Song, Wenbo Gao, Yuxiang Yang, Krzysztof Choromanski, Aldo Pacchiano, Yunhao Tang 
Publication date: 25 September 2019
Abstract: We introduce ES-MAML, a new framework for solving the model agnostic meta learning (MAML) problem based on Evolution Strategies (ES). Existing algorithms for MAML are based on policy gradients, and incur significant difficulties when attempting to estimate second derivatives using backpropagation on stochastic policies. We show how ES can be applied to MAML to obtain an algorithm which avoids the problem of estimating second derivatives, and is also conceptually simple and easy to implement. Moreover, ES-MAML can handle new types of nonsmooth adaptation operators, and other techniques for improving performance and estimation of ES methods become applicable. We show empirically that ES-MAML is competitive with existing methods and often yields better adaptation with fewer queries.
Has companion code repository: https://github.com/google-research/google-research/tree/master/es_maml
This page was built for publication: ES-MAML: Simple Hessian-Free Meta Learning
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6326439)
