Large-Scale Evolution of Image Classifiers

TL;DR

This paper demonstrates that evolutionary algorithms can automatically discover high-accuracy image classifiers for CIFAR datasets at large scales, reducing human intervention and emphasizing repeatability and computational considerations.

Contribution

It introduces a scalable evolutionary approach with novel mutation operators to automatically generate competitive image classification models without human design.

Findings

Achieved 94.6% accuracy on CIFAR-10

Achieved 77.0% accuracy on CIFAR-100

Demonstrated repeatability and manageable computational requirements

Abstract

Neural networks have proven effective at solving difficult problems but designing their architectures can be challenging, even for image classification problems alone. Our goal is to minimize human participation, so we employ evolutionary algorithms to discover such networks automatically. Despite significant computational requirements, we show that it is now possible to evolve models with accuracies within the range of those published in the last year. Specifically, we employ simple evolutionary techniques at unprecedented scales to discover models for the CIFAR-10 and CIFAR-100 datasets, starting from trivial initial conditions and reaching accuracies of 94.6% (95.6% for ensemble) and 77.0%, respectively. To do this, we use novel and intuitive mutation operators that navigate large search spaces; we stress that no human participation is required once evolution starts and that the output is a fully-trained model. Throughout this work, we place special emphasis on the repeatability of results, the variability in the outcomes and the computational requirements.