Handling Imbalanced Pseudolabels for Vision-Language Models with Concept Alignment and Confusion-Aware Calibrated Margin

TL;DR

This paper addresses the challenge of imbalanced pseudolabels in vision-language models by identifying key causes and proposing a novel framework with concept alignment and confusion-aware calibration, significantly improving label balance and accuracy.

Contribution

It introduces a new framework that mitigates pseudolabel imbalance by tackling concept mismatch and confusion, with mechanisms for concept alignment and calibrated margins.

Findings

Achieves a 6.29% relative improvement over state-of-the-art methods.

Effectively enhances pseudolabel accuracy and class balance across six datasets.

Demonstrates robustness across three learning paradigms.

Abstract

Adapting vision-language models (VLMs) to downstream tasks with pseudolabels has gained increasing attention. A major obstacle is that the pseudolabels generated by VLMs tend to be imbalanced, leading to inferior performance. While existing methods have explored various strategies to address this, the underlying causes of imbalance remain insufficiently investigated. To fill this gap, we delve into imbalanced pseudolabels and identify two primary contributing factors: concept mismatch and concept confusion. To mitigate these two issues, we propose a novel framework incorporating concept alignment and confusion-aware calibrated margin mechanisms. The core of our approach lies in enhancing underperforming classes and promoting balanced predictions across categories, thus mitigating imbalance. Extensive experiments on six benchmark datasets with three learning paradigms demonstrate that the proposed method effectively enhances the accuracy and balance of pseudolabels, achieving a relative improvement of 6.29% over the SoTA method. Our code is avaliable at https://anonymous.4open.science/r/CAP-C642/