Mixture of Balanced Information Bottlenecks for Long-Tailed Visual Recognition
Yifan Lan [email protected], Xin Cai [email protected], Jun Cheng [email protected], Shan Tan [email protected] · 2025
Read Mixture of Balanced Information Bottlenecks for Long-Tailed Visual Recognition by Yifan Lan [email protected], Xin Cai [email protected], Jun Cheng [email protected], Shan Tan [email protected] (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-2509-01804.txt.
Description
Yifan Lan, Xin Cai, Jun Cheng, and Shan Tan tackle image sorting when some classes are rare and others are common. Such long-tailed data, where a few classes dominate, trips up deep networks trained on it. Their tool is the information bottleneck, a method that squeezes inputs while keeping label relevant detail. They add loss rebalancing and self teaching steps to protect rare classes.
A mixture version runs several bottlenecks that draw on different network layers at once. Training then learns features and sorting together from an information theory view. Tests on three standard long-tailed sets, including a large nature photo set, reach top scores. Both the single and mixture versions beat prior methods in their checks.
Catalog record
| CREATOR | Yifan Lan [email protected], Xin Cai [email protected], Jun Cheng [email protected], Shan Tan [email protected] |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 57 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-01804