Procedural Generation of Algorithm Discovery Tasks in Machine Learning

TL;DR

DiscoGen is a procedural generator that creates a vast and diverse set of machine learning algorithm discovery tasks, enabling better evaluation and development of algorithm discovery agents.

Contribution

We introduce DiscoGen, a novel procedural generation framework for diverse algorithm discovery tasks, along with DiscoBench for standardized evaluation.

Findings

DiscoGen spans millions of tasks across machine learning fields.

DiscoBench provides a small, fixed subset for evaluation.

Experiments show DiscoGen can optimize algorithm discovery agents.

Abstract

Automating the development of machine learning algorithms has the potential to unlock new breakthroughs. However, our ability to improve and evaluate algorithm discovery systems has thus far been limited by existing task suites. They suffer from many issues, such as: poor evaluation methodologies; data contamination; and containing saturated or very similar problems. Here, we introduce DiscoGen, a procedural generator of algorithm discovery tasks for machine learning, such as developing optimisers for reinforcement learning or loss functions for image classification. Motivated by the success of procedural generation in reinforcement learning, DiscoGen spans millions of tasks of varying difficulty and complexity from a range of machine learning fields. These tasks are specified by a small number of configuration parameters and can be used to optimise algorithm discovery agents (ADAs). We present DiscoBench, a benchmark consisting of a fixed, small subset of DiscoGen tasks for principled evaluation of ADAs. Finally, we propose a number of ambitious, impactful research directions enabled by DiscoGen, in addition to experiments demonstrating its use for prompt optimisation of an ADA. DiscoGen is released open-source at https://github.com/AlexGoldie/discogen.