EchoAlign: Bridging Generative and Discriminative Learning under Noisy Labels

TL;DR

EchoAlign is a novel framework that enhances learning under noisy labels by combining generative and discriminative methods, improving robustness and accuracy in challenging noisy environments.

Contribution

It introduces a dual-component approach, EchoMod and EchoSelect, to adjust instance features and select reliable samples without label correction, advancing noisy label learning.

Findings

Outperforms state-of-the-art methods on benchmark datasets.

Retains nearly twice as many correctly labeled samples under 30% noise.

Maintains 99% selection accuracy in noisy settings.

Abstract

Noisy labels severely hinder the accuracy and generalization of machine learning models, especially when ambiguous instance features make reliable annotation difficult. Existing approaches, including transition-matrix-based label correction, struggle to capture complex relationships between instances and noisy labels, limiting their effectiveness in such settings. We present EchoAlign, a framework that bridges generative and discriminative learning under noisy labels. Instead of correcting labels, EchoAlign treats noisy labels as supervision targets and modifies the corresponding instances to align with them. The framework has two components: EchoMod uses controllable generative models to adjust instance features while preserving key instance-level structural cues, such as shape and edges, and avoiding excessive distortion; EchoSelect mitigates distribution shifts by retaining a reliable subset of original instances, guided by feature similarity between original and modified samples. This generative-discriminative interplay enables robust learning in highly noisy settings. Experiments on three benchmark datasets show that EchoAlign outperforms state-of-the-art methods in most evaluated settings. Under 30% instance-dependent noise, EchoSelect retains nearly twice as many correctly labeled samples as competing approaches while maintaining 99% selection accuracy, demonstrating the robustness and effectiveness of EchoAlign.