Study Diagnoses Prototype Estimation Errors in Few-Shot Class-Incremental Learning

A new arXiv paper examines why prototype heads with frozen visual encoders still fall short in few-shot class-incremental learning, proposing training-free geometry selection and unlabeled-query prototype rectification.

Read the full story at arXiv cs.CV ↗

Timeline · 1 report

1. 2026-11-03 05:00 · arXiv cs.CV
   Diagnosing Prototype Estimation in Few-Shot Class-Incremental Learning: Training-Free Geometry Selection and Unlabeled-Query Prototype Rectification