CARE: Class-Adaptive Expert Consensus for Reliable Learning with Long-Tailed Noisy Labels

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Mengke, Ling, Haiquan, Chen, Lihao, Lu, Yang, Zhang, Yiqun, Huang, Hui
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916038389530624
author Li, Mengke
Ling, Haiquan
Chen, Lihao
Lu, Yang
Zhang, Yiqun
Huang, Hui
author_facet Li, Mengke
Ling, Haiquan
Chen, Lihao
Lu, Yang
Zhang, Yiqun
Huang, Hui
contents Learning from real-world data is frequently hindered by the compound challenge of long-tailed class distributions and noisy annotations. Existing methods partially address these issues but typically ignore the non-uniform impact of label noise across classes, resulting in ineffective correction for tail classes and over-regularization for head classes. To address this issue, we propose Class-Adaptive Rectification with Experts (CARE), a parameter-efficient framework that leverages three complementary supervision sources from vision-language models (VLM): observed noisy labels, VLM text embeddings, and visual features. CARE introduces a class-adaptive expert consensus mechanism that enforces stricter agreement for tail classes and more permissive agreement for head classes based on class frequency. By aggregating high-confidence predictions across these sources, CARE filters unreliable signals and recalibrates class distributions, yielding more reliable rectification under long-tailed distributions. Extensive experiments on both synthetic and real-world benchmarks demonstrate that CARE consistently outperforms state-of-the-art methods, achieving up to 3.0\% performance gains. The source code is available at https://github.com/qwq123-study/CARE.
format Preprint
id arxiv_https___arxiv_org_abs_2605_23254
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle CARE: Class-Adaptive Expert Consensus for Reliable Learning with Long-Tailed Noisy Labels
Li, Mengke
Ling, Haiquan
Chen, Lihao
Lu, Yang
Zhang, Yiqun
Huang, Hui
Computer Vision and Pattern Recognition
Learning from real-world data is frequently hindered by the compound challenge of long-tailed class distributions and noisy annotations. Existing methods partially address these issues but typically ignore the non-uniform impact of label noise across classes, resulting in ineffective correction for tail classes and over-regularization for head classes. To address this issue, we propose Class-Adaptive Rectification with Experts (CARE), a parameter-efficient framework that leverages three complementary supervision sources from vision-language models (VLM): observed noisy labels, VLM text embeddings, and visual features. CARE introduces a class-adaptive expert consensus mechanism that enforces stricter agreement for tail classes and more permissive agreement for head classes based on class frequency. By aggregating high-confidence predictions across these sources, CARE filters unreliable signals and recalibrates class distributions, yielding more reliable rectification under long-tailed distributions. Extensive experiments on both synthetic and real-world benchmarks demonstrate that CARE consistently outperforms state-of-the-art methods, achieving up to 3.0\% performance gains. The source code is available at https://github.com/qwq123-study/CARE.
title CARE: Class-Adaptive Expert Consensus for Reliable Learning with Long-Tailed Noisy Labels
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.23254