![[2602.22297] Learning Rewards, Not Labels: Adversarial Inverse Reinforcement Learning for Machinery Fault Detection](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.22297] Learning Rewards, Not Labels: Adversarial Inverse Reinforcement Learning for Machinery Fault Detection
Summary
This paper presents a novel approach to machinery fault detection using Adversarial Inverse Reinforcement Learning, enabling effective anomaly detection without manual reward labels.
Why It Matters
The research addresses the limitations of traditional reinforcement learning methods in machinery fault detection by leveraging inverse reinforcement learning. This approach enhances the accuracy and efficiency of fault detection systems, which is critical in industrial settings for minimizing downtime and maintenance costs.
Key Takeaways
- Proposes a new framework for machinery fault detection using Adversarial Inverse Reinforcement Learning.
- Eliminates the need for manual reward engineering by learning reward dynamics from healthy operational sequences.
- Demonstrates effectiveness on benchmark datasets with consistent anomaly detection performance.
- Aligns reinforcement learning's sequential reasoning with the temporal structure of fault detection.
- Offers potential for improved diagnostics in data-driven industrial applications.
Computer Science > Machine Learning arXiv:2602.22297 (cs) [Submitted on 25 Feb 2026] Title:Learning Rewards, Not Labels: Adversarial Inverse Reinforcement Learning for Machinery Fault Detection Authors:Dhiraj Neupane, Richard Dazeley, Mohamed Reda Bouadjenek, Sunil Aryal View a PDF of the paper titled Learning Rewards, Not Labels: Adversarial Inverse Reinforcement Learning for Machinery Fault Detection, by Dhiraj Neupane and 3 other authors View PDF HTML (experimental) Abstract:Reinforcement learning (RL) offers significant promise for machinery fault detection (MFD). However, most existing RL-based MFD approaches do not fully exploit RL's sequential decision-making strengths, often treating MFD as a simple guessing game (Contextual Bandits). To bridge this gap, we formulate MFD as an offline inverse reinforcement learning problem, where the agent learns the reward dynamics directly from healthy operational sequences, thereby bypassing the need for manual reward engineering and fault labels. Our framework employs Adversarial Inverse Reinforcement Learning to train a discriminator that distinguishes between normal (expert) and policy-generated transitions. The discriminator's learned reward serves as an anomaly score, indicating deviations from normal operating behaviour. When evaluated on three run-to-failure benchmark datasets (HUMS2023, IMS, and XJTU-SY), the model consistently assigns low anomaly scores to normal samples and high scores to faulty ones, enabling early an...
