AutoML-Zero: Evolving Machine Learning Algorithms From Scratch

TL;DR

This paper demonstrates that AutoML can automatically discover complete machine learning algorithms from basic mathematical operations, leading to the emergence of modern techniques through evolutionary search.

Contribution

It introduces a novel framework that reduces human bias and enables the automatic evolution of entire machine learning algorithms from scratch.

Findings

Evolutionary search can discover simple neural networks trained by backpropagation.

Evolved algorithms can outperform baseline models on tasks like CIFAR-10.

Task-specific adaptations, such as dropout-like techniques, emerge naturally.

Abstract

Machine learning research has advanced in multiple aspects, including model structures and learning methods. The effort to automate such research, known as AutoML, has also made significant progress. However, this progress has largely focused on the architecture of neural networks, where it has relied on sophisticated expert-designed layers as building blocks---or similarly restrictive search spaces. Our goal is to show that AutoML can go further: it is possible today to automatically discover complete machine learning algorithms just using basic mathematical operations as building blocks. We demonstrate this by introducing a novel framework that significantly reduces human bias through a generic search space. Despite the vastness of this space, evolutionary search can still discover two-layer neural networks trained by backpropagation. These simple neural networks can then be surpassed by evolving directly on tasks of interest, e.g. CIFAR-10 variants, where modern techniques emerge in the top algorithms, such as bilinear interactions, normalized gradients, and weight averaging. Moreover, evolution adapts algorithms to different task types: e.g., dropout-like techniques appear when little data is available. We believe these preliminary successes in discovering machine learning algorithms from scratch indicate a promising new direction for the field.