arXiv：cs.LG· Aleix Alcacer, Irene Epifanio, Sebastian Mair, Morten M{\o}rup·· 3 小时前
原型分析（Archetypal Analysis）综述
A Survey on Archetypal Analysis
AI 导读
这是首篇系统综述原型分析（AA）的论文，AA 由 Adele Cutler 和 Leo Breiman 于 1994 年提出，通过凸组合将观测记录近似表示为若干原型，实现可解释的特征提取与降维。该综述梳理了 AA 在多个科学领域的应用、建模最佳实践及其非凸优化等局限，并指出未来研究方向，已被 TPAMI 2026 接收。
正文
Abstract:Archetypal analysis (AA) was originally proposed in 1994 by Adele Cutler and Leo Breiman as a computational procedure for extracting distinct aspects, so-called archetypes, from observations, with each observational record approximated as a mixture (i.e., convex combination) of these archetypes. AA thereby provides straightforward, interpretable, and explainable representations for feature extraction and dimensionality reduction, facilitating the understanding of the structure of high-dimensional data and enabling wide applications across the sciences. However, AA also faces challenges, particularly as the associated optimization problem is nonconvex. This is the first survey that provides researchers and data mining practitioners with an overview of the methodologies and opportunities that AA offers, surveying the many applications of AA across disparate fields of science, as well as best practices for modeling data with AA and its limitations. The survey concludes by explaining crucial future research directions concerning AA.
| Comments: | 28 pages, 14 figures, accepted at TPAMI |
| Subjects: | Methodology (stat.ME); Machine Learning (cs.LG); Machine Learning (stat.ML) |
| Cite as: | arXiv:2504.12392 [stat.ME] |
| (or arXiv:2504.12392v3 [stat.ME] for this version) | |
| https://doi.org/10.48550/arXiv.2504.12392 arXiv-issued DOI via DataCite |
|
| Journal reference: | IEEE Transactions on Pattern Analysis and Machine Intelligence 2026 |
| Related DOI: | https://doi.org/10.1109/TPAMI.2026.3740133
DOI(s) linking to related resources |
Submission history
From: Sebastian Mair [view email]
[v1]
Wed, 16 Apr 2025 18:01:05 UTC (3,566 KB)
[v2]
Fri, 19 Dec 2025 12:20:38 UTC (4,756 KB)
[v3]
Wed, 7 Oct 2026 19:49:25 UTC (4,763 KB)
来源：arXiv：cs.LG · arxiv.org