![[2508.16832] Out of Distribution Detection for Efficient Continual Learning in Quality Prediction for Arc Welding](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2508.16832] Out of Distribution Detection for Efficient Continual Learning in Quality Prediction for Arc Welding
Summary
This article presents a novel approach to out-of-distribution detection in arc welding quality prediction, enhancing continual learning by minimizing labeling needs and adapting to dynamic manufacturing environments.
Why It Matters
The research addresses critical challenges in manufacturing where machine learning models often fail due to distribution shifts. By improving out-of-distribution detection, the study contributes to more robust AI systems, essential for quality assurance in dynamic industrial processes.
Key Takeaways
- Introduces an enhanced VQ-VAE Transformer for quality prediction in welding.
- Integrates out-of-distribution detection with continual learning to optimize model updates.
- Demonstrates superior performance over traditional methods in real-world scenarios.
- Proposes a new quantitative metric for evaluating OOD detection and in-distribution performance.
- Addresses the need for adaptive AI solutions in dynamic manufacturing environments.
Computer Science > Machine Learning arXiv:2508.16832 (cs) [Submitted on 22 Aug 2025 (v1), last revised 16 Feb 2026 (this version, v2)] Title:Out of Distribution Detection for Efficient Continual Learning in Quality Prediction for Arc Welding Authors:Yannik Hahn, Jan Voets, Antonin Koenigsfeld, Hasan Tercan, Tobias Meisen View a PDF of the paper titled Out of Distribution Detection for Efficient Continual Learning in Quality Prediction for Arc Welding, by Yannik Hahn and 4 other authors View PDF HTML (experimental) Abstract:Modern manufacturing relies heavily on fusion welding processes, including gas metal arc welding (GMAW). Despite significant advances in machine learning-based quality prediction, current models exhibit critical limitations when confronted with the inherent distribution shifts that occur in dynamic manufacturing environments. In this work, we extend the VQ-VAE Transformer architecture - previously demonstrating state-of-the-art performance in weld quality prediction - by leveraging its autoregressive loss as a reliable out-of-distribution (OOD) detection mechanism. Our approach exhibits superior performance compared to conventional reconstruction methods, embedding error-based techniques, and other established baselines. By integrating OOD detection with continual learning strategies, we optimize model adaptation, triggering updates only when necessary and thereby minimizing costly labeling requirements. We introduce a novel quantitative metric that simu...
