Learning to Optimize Neural Nets
This work addresses the challenge of optimizing neural networks for researchers and practitioners, but it is incremental as it extends an existing framework to a specific high-dimensional setting.
The paper tackles the problem of learning optimization algorithms for training shallow neural nets using reinforcement learning, demonstrating that the learned algorithm outperforms existing methods on unseen tasks and is robust to changes in gradient stochasticity and architecture, with generalization shown across datasets like MNIST, Toronto Faces, CIFAR-10, and CIFAR-100.
Learning to Optimize is a recently proposed framework for learning optimization algorithms using reinforcement learning. In this paper, we explore learning an optimization algorithm for training shallow neural nets. Such high-dimensional stochastic optimization problems present interesting challenges for existing reinforcement learning algorithms. We develop an extension that is suited to learning optimization algorithms in this setting and demonstrate that the learned optimization algorithm consistently outperforms other known optimization algorithms even on unseen tasks and is robust to changes in stochasticity of gradients and the neural net architecture. More specifically, we show that an optimization algorithm trained with the proposed method on the problem of training a neural net on MNIST generalizes to the problems of training neural nets on the Toronto Faces Dataset, CIFAR-10 and CIFAR-100.