Research map / Label-efficient and robust learning / Recognition and long-tail
TailCon: Mitigating Tail Signal Erosion through Memory Consolidation for Long-Tailed Recognition
Yuan Dong, Di Wu, Zhe Zhao, Liheng Yu, Xiaofeng Cao, Pengkun Wang, Yang Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes TailCon, a memory-consolidation approach to counter tail signal erosion, where frequent-class updates dilute sparse rare-class signals in long-tailed recognition.
Problem
Long-tailed recognition is hard because rare classes have limited supervision and their evidence erodes during standard stochastic training when frequent-class updates dilute sparse rare-class signals.
Approach
Characterizes the failure as tail signal erosion in conflict-dominant regimes and proposes a memory consolidation mechanism (TailCon) to let rare-class evidence accumulate before it is overwritten by frequent-class updates.
Closest papers by meaning
- A Theoretical Bridge Between Long-Tailed Recognition and Continual Learning (NeurIPS 2026 · similarity 0.90)
- Echoes of the Past: Boosting Long-tail Recognition via Reflective Learning (ECCV 2024 · similarity 0.88)
- Feature Fusion from Head to Tail for Long-Tailed Visual Recognition (AAAI 2024 · similarity 0.87)
- Long-tailed Recognition with Model Rebalancing (NeurIPS 2025 · similarity 0.87)
- DiffuLT: Diffusion for Long-tail Recognition Without External Knowledge (NeurIPS 2024 · similarity 0.87)
- Decoupled Optimisation for Long-Tailed Visual Recognition (AAAI 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
