![[2506.05647] Learning to Weight Parameters for Training Data Attribution](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2506.05647] Learning to Weight Parameters for Training Data Attribution
Summary
This paper introduces a novel method for gradient-based data attribution that learns parameter importance weights from data, enhancing attribution accuracy across various tasks without requiring annotated labels.
Why It Matters
Understanding which training examples influence model outputs is crucial for improving model transparency and performance. This research addresses limitations in existing methods by providing a more accurate and data-driven approach to parameter weighting, which can enhance applications in machine learning and AI.
Key Takeaways
- Proposes a method to learn parameter importance weights directly from data.
- Improves attribution accuracy for diverse tasks, including image classification and language modeling.
- Eliminates the need for annotated labels in the attribution process.
- Addresses functional heterogeneity of network parameters more effectively than existing methods.
- Supports fine-grained attribution for specific concepts like subject and style.
Computer Science > Machine Learning arXiv:2506.05647 (cs) [Submitted on 6 Jun 2025 (v1), last revised 20 Feb 2026 (this version, v4)] Title:Learning to Weight Parameters for Training Data Attribution Authors:Shuangqi Li, Hieu Le, Jingyi Xu, Mathieu Salzmann View a PDF of the paper titled Learning to Weight Parameters for Training Data Attribution, by Shuangqi Li and 3 other authors View PDF HTML (experimental) Abstract:We study gradient-based data attribution, aiming to identify which training examples most influence a given output. Existing methods for this task either treat network parameters uniformly or rely on implicit weighting derived from Hessian approximations, which do not fully model functional heterogeneity of network parameters. To address this, we propose a method to explicitly learn parameter importance weights directly from data, without requiring annotated labels. Our approach improves attribution accuracy across diverse tasks, including image classification, language modeling, and diffusion, and enables fine-grained attribution for concepts like subject and style. Comments: Subjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2506.05647 [cs.LG] (or arXiv:2506.05647v4 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2506.05647 Focus to learn more arXiv-issued DOI via DataCite Journal reference: ICLR 2026 Submission history From: Shuangqi Li [view email] [v1] Fri, 6 Jun 2025 00:32:04 UTC (956 KB) [v2...
