Credal Machine Learning for Risk-Averse Decision Making

2026-10-08

AI Summary

This study addresses the unreliability of traditional risk-averse models caused by uncertainty in loss distributions. To mitigate this, the proposed method employs credal sets to represent epistemic uncertainty and introduces an efficient learner based on a set of probability distributions. Furthermore, a novel decision rule is designed to map a distribution set into a single distribution, thereby optimizing the conditional value-at-risk (CVaR) minimization problem. Applicable to both classification and reinforcement learning tasks, this approach effectively prevents catastrophic decisions under scenarios such as distribution shift while incurring negligible sacrifice in expected performance. Overall, the method significantly enhances the robustness and reliability of risk-averse predictions.

Abstract

In many machine learning applications, it is necessary to guard against worst-case scenarios and predictions that could result in substantial losses. In principle, this can be achieved by training risk-averse predictive models that minimize loss functions such as conditional value-at-risk (CVaR), rather than relying on models that perform well on average. In practice, however, the effectiveness of this approach to risk aversion is undermined by the learner's uncertainty regarding the true loss distribution and, consequently, the true CVaR. To achieve reliable risk-aversion, we propose a method in which this (epistemic) uncertainty is represented in terms of credal sets, i.e., sets of probability distributions. More specifically, we develop an efficient yet reliable learner that produces predictions in the form of credal sets and combine it with a novel decision rule that maps each credal set to a single predictive distribution for CVaR minimization. Across classification, under distribution shift, and in reinforcement learning, our approach reliably avoids catastrophic decisions, while sacrificing little in expected performance.

Problem

• Risk-averse decision making
• Conditional value-at-risk
• Epistemic uncertainty
• Credal sets
• Machine learning

Innovation

• Credal Sets
• Risk-Averse Decision Making
• Conditional Value-at-Risk (CVaR)
• Epistemic Uncertainty
• Distribution Shift