Full Collapse Is Suboptimal: Controlled Partial Neural Collapse for Long-Tailed Recognition
arXiv:2610.36460v1 Announce Type: new Abstract: Recent long-tailed recognition (LTR) methods incorporate neural collapse (NC) and align features to a fixed Equiangular Tight Frame (ETF) as a uniform target for all classes. However, forcing tail classes to fully collapse over-contracts their samples…
Read the full story at arXiv cs.CV ↗
Timeline · 1 report
- 2026-10-28 04:00 · arXiv cs.CV
Full Collapse Is Suboptimal: Controlled Partial Neural Collapse for Long-Tailed Recognition