Saved in:
Bibliographic Details
Main Authors: Conserva, Michelangelo, Sasso, Remo, Rauber, Paulo
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.17092
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912597079490560
author Conserva, Michelangelo
Sasso, Remo
Rauber, Paulo
author_facet Conserva, Michelangelo
Sasso, Remo
Rauber, Paulo
contents Principled evaluation is critical for progress in deep reinforcement learning (RL), yet it lags behind the theory-driven benchmarks of tabular RL. While tabular settings benefit from well-understood hardness measures like MDP diameter and suboptimality gaps, deep RL benchmarks are often chosen based on intuition and popularity. This raises a critical question: can tabular hardness metrics be adapted to guide non-tabular benchmarking? We investigate this question and reveal a fundamental gap. Our primary contribution is demonstrating that the difficulty of non-tabular environments is dominated by a factor that tabular metrics ignore: representation hardness. The same underlying MDP can pose vastly different challenges depending on whether the agent receives state vectors or pixel-based observations. To enable this analysis, we introduce \texttt{pharos}, a new open-source library for principled RL benchmarking that allows for systematic control over both environment structure and agent representations. Our extensive case study using \texttt{pharos} shows that while tabular metrics offer some insight, they are poor predictors of deep RL agent performance on their own. This work highlights the urgent need for new, representation-aware hardness measures and positions \texttt{pharos} as a key tool for developing them.
format Preprint
id arxiv_https___arxiv_org_abs_2509_17092
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle On the Limits of Tabular Hardness Metrics for Deep RL: A Study with the Pharos Benchmark
Conserva, Michelangelo
Sasso, Remo
Rauber, Paulo
Machine Learning
Principled evaluation is critical for progress in deep reinforcement learning (RL), yet it lags behind the theory-driven benchmarks of tabular RL. While tabular settings benefit from well-understood hardness measures like MDP diameter and suboptimality gaps, deep RL benchmarks are often chosen based on intuition and popularity. This raises a critical question: can tabular hardness metrics be adapted to guide non-tabular benchmarking? We investigate this question and reveal a fundamental gap. Our primary contribution is demonstrating that the difficulty of non-tabular environments is dominated by a factor that tabular metrics ignore: representation hardness. The same underlying MDP can pose vastly different challenges depending on whether the agent receives state vectors or pixel-based observations. To enable this analysis, we introduce \texttt{pharos}, a new open-source library for principled RL benchmarking that allows for systematic control over both environment structure and agent representations. Our extensive case study using \texttt{pharos} shows that while tabular metrics offer some insight, they are poor predictors of deep RL agent performance on their own. This work highlights the urgent need for new, representation-aware hardness measures and positions \texttt{pharos} as a key tool for developing them.
title On the Limits of Tabular Hardness Metrics for Deep RL: A Study with the Pharos Benchmark
topic Machine Learning
url https://arxiv.org/abs/2509.17092