# Distributionally Robust Inverse Reinforcement Learning for Identifying Multi-Agent Coordinated Sensing

arXiv: https://arxiv.org/abs/2409.14542 · HTML: https://mlsanity.com/paper/2409.14542/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Inverse reinforcement learning, one of 13 directions in Offline RL and imitation learning from data.
- Builds on 9 of its 21 references that are in the map.
- First posted on arXiv 22 Sep 2024.

## Placed in

- Offline RL and imitation learning from data / [Inverse reinforcement learning](https://mlsanity.com/offlinerl/inverse-reinforcement-learning/)

## Builds on (most cited first, 9 of 9)

- [Multi-Agent Adversarial Inverse Reinforcement Learning](https://mlsanity.com/paper/1907.13220/) (2019)
- [Identifying Cognitive Radars -- Inverse Reinforcement Learning using Revealed Preferences](https://mlsanity.com/paper/1912.00331/) (2019)
- [Adversarial Radar Inference: Inverse Tracking, Identifying Cognition and Designing Smart Interference](https://mlsanity.com/paper/2008.01559/) (2020)
- [Meta-Cognition. An Inverse-Inverse Reinforcement Learning Approach for Cognitive Radars](https://mlsanity.com/paper/2205.01794/) (2022)
- [Decomposition Algorithm for Distributionally Robust Optimization using Wasserstein Metric](https://mlsanity.com/paper/1704.03920/) (2017)
- [Wasserstein Distributionally Robust Inverse Multiobjective Optimization](https://mlsanity.com/paper/2009.14552/) (2020)
- [Identifying Coordination in a Cognitive Radar Network -- A Multi-Objective Inverse Reinforcement Learning Approach](https://mlsanity.com/paper/2211.06967/) (2022)
- [Statistical Detection of Coordination in a Cognitive Radar Network through Inverse Multi-objective Optimization](https://mlsanity.com/paper/2304.09125/) (2023)
- [Data-Driven Mechanism Design via Multi-Agent Revealed Preferences](https://mlsanity.com/paper/2404.15391/) (2024)

## Abstract (arXiv)

We derive a minimax distributionally robust inverse reinforcement learning (IRL) algorithm to reconstruct the utility functions of a multi-agent sensing system. Specifically, we construct utility estimators which minimize the worst-case prediction error over a Wasserstein ambiguity set centered at noisy signal observations. We prove the equivalence between this robust estimation and a semi-infinite optimization reformulation, and we propose a consistent algorithm to compute solutions. We illustrate the efficacy of this robust IRL scheme in numerical studies to reconstruct the utility functions of a cognitive radar network from observed tracking signals.