Holistic Reliability Propagation: Decoupling Annotation and Prediction for Robust Noisy-Label

TL;DR

This paper introduces a novel method called Holistic Reliability Propagation (HRP) that disentangles and estimates separate reliabilities for annotations and predictions, improving robustness in noisy-label multimedia classification.

Contribution

The paper proposes a bilevel meta-learning approach to estimate separate reliabilities for labels and pseudo-labels, enabling more effective noise handling in multimedia classification.

Findings

HRP improves average accuracy over strong baselines on synthetic and real-world benchmarks.

HRP remains competitive at high noise rates.

Disentangled reliabilities enhance robustness in noisy-label scenarios.

Abstract

Learning with noisy labels in multimedia classification often combines external annotations and model predictions into a single reliability weight, even though the two sources can fail for different reasons. We instead estimate disentangled reliabilities: bilevel meta-learning produces two batch-normalized scalars per sample, alpha for the given label and beta for the pseudo-label, without constraining them to sum to one. Holistic Reliability Propagation (HRP) then routes them to different objectives, using reliability-aware Mixup with global gating on the input branch and beta-gated pseudo-label positives on the contrastive branch. On synthetic and real-world benchmarks, HRP improves average accuracy over strong baselines and remains competitive at the highest noise rates.