Robust Reinforcement Learning with Dynamic Distortion Risk Measures

ArXiv ID: 2409.10096 “View on arXiv”

Authors: Unknown

Abstract

In a reinforcement learning (RL) setting, the agent’s optimal strategy heavily depends on her risk preferences and the underlying model dynamics of the training environment. These two aspects influence the agent’s ability to make well-informed and time-consistent decisions when facing testing environments. In this work, we devise a framework to solve robust risk-aware RL problems where we simultaneously account for environmental uncertainty and risk with a class of dynamic robust distortion risk measures. Robustness is introduced by considering all models within a Wasserstein ball around a reference model. We estimate such dynamic robust risk measures using neural networks by making use of strictly consistent scoring functions, derive policy gradient formulae using the quantile representation of distortion risk measures, and construct an actor-critic algorithm to solve this class of robust risk-aware RL problems. We demonstrate the performance of our algorithm on a portfolio allocation example.

Keywords: reinforcement learning, Wasserstein ball, distortion risk measures, portfolio allocation, robust risk-aware RL, Multi-Asset

Complexity vs Empirical Score

• Math Complexity: 8.5/10
• Empirical Rigor: 4.0/10
• Quadrant: Lab Rats
• Why: The paper is dense with advanced mathematics, including Wasserstein distance, dynamic distortion risk measures, quantile representations, and policy gradient derivations, but the empirical component is limited to a single portfolio allocation example without code or extensive statistical validation.

  flowchart TD
    A["Research Goal: Robust Risk-Aware RL"] --> B["Methodology: Dynamic Robust Distortion Risk Measures"]
    B --> C{"Data: Portfolio Allocation Example"}
    C --> D["Computation: Policy Gradient via Quantile Representation"]
    D --> E["Computation: Actor-Critic Algorithm"]
    E --> F["Outcome: Robust Strategy Optimization"]
    F --> G["Key Finding: Performance on Multi-Asset Portfolio"]