Unsupervised Learning for Class Distribution Mismatch

TL;DR

This paper introduces UCDM, an unsupervised method that effectively handles class distribution mismatch by synthesizing training pairs and iteratively pseudo-labeling unlabeled data, outperforming semi-supervised methods.

Contribution

The paper presents a novel unsupervised approach for class distribution mismatch that does not rely on labeled data, using data synthesis and confidence-based pseudo-labeling.

Findings

UCDM outperforms previous semi-supervised methods on multiple datasets.

Achieves significant accuracy improvements with 60% class mismatch.

Effectively classifies known, unknown, and new classes without labeled data.

Abstract

Class distribution mismatch (CDM) refers to the discrepancy between class distributions in training data and target tasks. Previous methods address this by designing classifiers to categorize classes known during training, while grouping unknown or new classes into an "other" category. However, they focus on semi-supervised scenarios and heavily rely on labeled data, limiting their applicability and performance. To address this, we propose Unsupervised Learning for Class Distribution Mismatch (UCDM), which constructs positive-negative pairs from unlabeled data for classifier training. Our approach randomly samples images and uses a diffusion model to add or erase semantic classes, synthesizing diverse training pairs. Additionally, we introduce a confidence-based labeling mechanism that iteratively assigns pseudo-labels to valuable real-world data and incorporates them into the training process. Extensive experiments on three datasets demonstrate UCDM's superiority over previous semi-supervised methods. Specifically, with a 60% mismatch proportion on Tiny-ImageNet dataset, our approach, without relying on labeled data, surpasses OpenMatch (with 40 labels per class) by 35.1%, 63.7%, and 72.5% in classifying known, unknown, and new classes.