Evolutionary Stochastic Gradient Descent for Optimization of Deep Neural Networks

Abstract

We propose a population-based Evolutionary Stochastic Gradient Descent (ESGD) framework for optimizing deep neural networks. ESGD combines SGD and gradient-free evolutionary algorithms as complementary algorithms in one framework in which the optimization alternates between the SGD step and evolution step to improve the average fitness of the population. With a back-off strategy in the SGD step and an elitist strategy in the evolution step, it guarantees that the best fitness in the population will never degrade. In addition, individuals in the population optimized with various SGD-based optimizers using distinct hyper-parameters in the SGD step are considered as competing species in a coevolution setting such that the complementarity of the optimizers is also taken into account. The effectiveness of ESGD is demonstrated across multiple applications including speech recognition, image recognition and language modeling, using networks with a variety of deep architectures.

Cited by in corpus (6)

• Autoencoding with a Classifier System
• Mixing ADAM and SGD: a Combined Optimization Method
• Improving Gradient Estimation in Evolutionary Strategies With Past Descent Directions
• Accelerating Reinforcement Learning with a Directional-Gaussian-Smoothing Evolution Strategy
• Unbiased Gradient Estimation in Unrolled Computation Graphs with Persistent Evolution Strategies
• Acoustic Model Optimization Based On Evolutionary Stochastic Gradient Descent with Anchors for Automatic Speech Recognition