How Many Random Seeds? Statistical Power Analysis in Deep Reinforcement Learning Experiments

TL;DR

This paper discusses how to determine the appropriate number of random seeds in deep reinforcement learning experiments to ensure statistically valid comparisons, addressing reproducibility concerns.

Contribution

It provides theoretical guidelines for selecting the number of random seeds using t-tests and bootstrap methods, and discusses the impact of assumption deviations on statistical evaluations.

Findings

Guidelines for choosing the number of random seeds for significance testing

Analysis of how assumption violations affect statistical error estimates

Provision of code for conducting the recommended tests

Abstract

Consistently checking the statistical significance of experimental results is one of the mandatory methodological steps to address the so-called "reproducibility crisis" in deep reinforcement learning. In this tutorial paper, we explain how the number of random seeds relates to the probabilities of statistical errors. For both the t-test and the bootstrap confidence interval test, we recall theoretical guidelines to determine the number of random seeds one should use to provide a statistically significant comparison of the performance of two algorithms. Finally, we discuss the influence of deviations from the assumptions usually made by statistical tests. We show that they can lead to inaccurate evaluations of statistical errors and provide guidelines to counter these negative effects. We make our code available to perform the tests.