Neural Optimizer Search with Reinforcement Learning

TL;DR

This paper introduces a reinforcement learning-based method to automatically discover and optimize neural network update rules, resulting in new optimizers that outperform traditional methods across various tasks.

Contribution

It presents a novel neural optimizer search framework using reinforcement learning, leading to the discovery of effective new optimization algorithms for deep learning.

Findings

Discovered update rules outperform Adam, RMSProp, SGD on CIFAR-10

Introduced two new optimizers: PowerSign and AddSign

New optimizers transfer well to other tasks like ImageNet and machine translation

Abstract

We present an approach to automate the process of discovering optimization methods, with a focus on deep learning architectures. We train a Recurrent Neural Network controller to generate a string in a domain specific language that describes a mathematical update equation based on a list of primitive functions, such as the gradient, running average of the gradient, etc. The controller is trained with Reinforcement Learning to maximize the performance of a model after a few epochs. On CIFAR-10, our method discovers several update rules that are better than many commonly used optimizers, such as Adam, RMSProp, or SGD with and without Momentum on a ConvNet model. We introduce two new optimizers, named PowerSign and AddSign, which we show transfer well and improve training on a variety of different tasks and architectures, including ImageNet classification and Google's neural machine translation system.