TailedCore: Few-Shot Sampling for Unsupervised Long-Tail Noisy Anomaly Detection
Yoon Gyo Jung, Jaewoo Park, Jaeho Yoon, Kuan-Chuan Peng, Wonchul Kim, Andrew Beng Jin Teoh, Octavia Camps · 2025
Read TailedCore: Few-Shot Sampling for Unsupervised Long-Tail Noisy Anomaly Detection by Yoon Gyo Jung, Jaewoo Park, Jaeho Yoon, Kuan-Chuan Peng, Wonchul Kim, Andrew Beng Jin Teoh, Octavia Camps (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-02775.txt.
Description
Jung and co-authors build a defect finder for hard factory data in a 2025 computer science paper. Real image sets are messy: normal samples hide flaws, and rare product types form a long tail. Models face a trade off, since guarding against noise hurts rare types and guarding rare types hurts noise handling. The team splits the two problems apart.
A new sampler guesses each class size from how alike the sample vectors are, under a symmetry assumption. That lets the system pull rare class samples aside and treat them on their own. The full model, TailedCore, stores a memory that keeps tail detail while shrugging off noise. Tests on the messy setting show the model beating prior best methods in most cases.
Catalog record
| CREATOR | Yoon Gyo Jung, Jaewoo Park, Jaeho Yoon, Kuan-Chuan Peng, Wonchul Kim, Andrew Beng Jin Teoh, Octavia Camps |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 92 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-02775