Is Data Shapley Not Better than Random in Data Selection? Ask NASH
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909035064721408 |
|---|---|
| author | Tian, Xiao Fan, Jue Sim, Rachael Hwee Ling Wang, Zixuan Chen, Nancy F. Low, Bryan Kian Hsiang |
| author_facet | Tian, Xiao Fan, Jue Sim, Rachael Hwee Ling Wang, Zixuan Chen, Nancy F. Low, Bryan Kian Hsiang |
| contents | Data selection studies the problem of identifying high-quality subsets of training data. While some existing works have considered selecting the subset of data with top-$m$ Data Shapley or other semivalues as they account for the interaction among every subset of data, other works argue that Data Shapley can sometimes perform ineffectively in practice and select subsets that are no better than random. This raises the questions: (I) Are there certain "Shapley-informative" settings where Data Shapley consistently works well? (II) Can we strategically utilize these settings to select high-quality subsets consistently and efficiently? In this paper, we propose a novel data selection framework, NASH (Non-linear Aggregation of SHapley-informative components), which (I) decomposes the target utility function (e.g., validation accuracy) into simpler, Shapley-informative component functions, and selects data by optimizing an objective that (II) aggregates these components non-linearly. We demonstrate that NASH substantially boosts the effectiveness of Shapley/semivalue-based data selection with minimal additional runtime cost. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_10684 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Is Data Shapley Not Better than Random in Data Selection? Ask NASH Tian, Xiao Fan, Jue Sim, Rachael Hwee Ling Wang, Zixuan Chen, Nancy F. Low, Bryan Kian Hsiang Machine Learning Artificial Intelligence Data selection studies the problem of identifying high-quality subsets of training data. While some existing works have considered selecting the subset of data with top-$m$ Data Shapley or other semivalues as they account for the interaction among every subset of data, other works argue that Data Shapley can sometimes perform ineffectively in practice and select subsets that are no better than random. This raises the questions: (I) Are there certain "Shapley-informative" settings where Data Shapley consistently works well? (II) Can we strategically utilize these settings to select high-quality subsets consistently and efficiently? In this paper, we propose a novel data selection framework, NASH (Non-linear Aggregation of SHapley-informative components), which (I) decomposes the target utility function (e.g., validation accuracy) into simpler, Shapley-informative component functions, and selects data by optimizing an objective that (II) aggregates these components non-linearly. We demonstrate that NASH substantially boosts the effectiveness of Shapley/semivalue-based data selection with minimal additional runtime cost. |
| title | Is Data Shapley Not Better than Random in Data Selection? Ask NASH |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2605.10684 |