pith. sign in

arxiv: 1812.02341 · v3 · pith:UMCPILPInew · submitted 2018-12-06 · 💻 cs.LG · stat.ML

Quantifying Generalization in Reinforcement Learning

classification 💻 cs.LG stat.ML
keywords generalizationlearningtrainingcoinrunenvironmentsreinforcementsetsability
0
0 comments X
read the original abstract

In this paper, we investigate the problem of overfitting in deep reinforcement learning. Among the most common benchmarks in RL, it is customary to use the same environments for both training and testing. This practice offers relatively little insight into an agent's ability to generalize. We address this issue by using procedurally generated environments to construct distinct training and test sets. Most notably, we introduce a new environment called CoinRun, designed as a benchmark for generalization in RL. Using CoinRun, we find that agents overfit to surprisingly large training sets. We then show that deeper convolutional architectures improve generalization, as do methods traditionally found in supervised learning, including L2 regularization, dropout, data augmentation and batch normalization.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Dota 2 with Large Scale Deep Reinforcement Learning

    cs.LG 2019-12 accept novelty 7.0

    OpenAI Five achieved superhuman performance in Dota 2 by defeating the world champions using scaled self-play reinforcement learning.

  2. Understanding Goal Generalisation in Sequential Reinforcement Learning

    cs.LG 2026-05 unverdicted novelty 6.0

    Empirical analysis of over 100 sequential RL training pipelines across 250+ OOD environments finds salient features drive generalization and early goals persist, with latent policy gradients simulating latent variable...

  3. Generalizing from a few environments in safety-critical reinforcement learning

    cs.LG 2019-07 unverdicted novelty 6.0

    RL agents fail dangerously on unseen environments; ensembles reduce catastrophes in gridworld but not CoinRun, with uncertainty enabling intervention prediction.

  4. Reasoning and Generalization in RL: A Tool Use Perspective

    cs.NE 2019-07 unverdicted novelty 5.0

    Proposes a tool-use inspired framework with multiple test sets to measure specified types of generalization in RL.