Translating the Rashomon Effect to Sequential Decision-Making Tasks

TL;DR

This paper explores the Rashomon effect in sequential decision-making, showing multiple policies can behave identically yet differ internally, and that ensembles from these policies enhance robustness and efficiency.

Contribution

It extends the Rashomon effect to sequential decision-making, introduces formal verification for policy comparison, and demonstrates benefits of Rashomon sets in robustness and computational efficiency.

Findings

Rashomon effect exists in sequential decision-making.

Ensembles from Rashomon set improve robustness to distribution shifts.

Permissive policies reduce verification complexity while maintaining performance.

Abstract

The Rashomon effect describes the phenomenon where multiple models trained on the same data produce identical predictions while differing in which features they rely on internally. This effect has been studied extensively in classification tasks, but not in sequential decision-making, where an agent learns a policy to achieve an objective by taking actions in an environment. In this paper, we translate the Rashomon effect to sequential decision-making. We define it as multiple policies that exhibit identical behavior, visiting the same states and selecting the same actions, while differing in their internal structure, such as feature attributions. Verifying identical behavior in sequential decision-making differs from classification. In classification, predictions can be directly compared to ground-truth labels. In sequential decision-making with stochastic transitions, the same policy may succeed or fail on any single trajectory due to randomness. We address this using formal verification methods that construct and compare the complete probabilistic behavior of each policy in the environment. Our experiments demonstrate that the Rashomon effect exists in sequential decision-making. We further show that ensembles constructed from the Rashomon set exhibit greater robustness to distribution shifts than individual policies. Additionally, permissive policies derived from the Rashomon set reduce computational requirements for verification while maintaining optimal performance.