Salvato in:
Dettagli Bibliografici
Autori principali: Du, Pan, Zhao, Wangbo, Lu, Xinai, Liu, Nian, Li, Zhikai, Gong, Chaoyu, Zhao, Suyun, Chen, Hong, Li, Cuiping, Wang, Kai, You, Yang
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2505.06948
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910937829605376
author Du, Pan
Zhao, Wangbo
Lu, Xinai
Liu, Nian
Li, Zhikai
Gong, Chaoyu
Zhao, Suyun
Chen, Hong
Li, Cuiping
Wang, Kai
You, Yang
author_facet Du, Pan
Zhao, Wangbo
Lu, Xinai
Liu, Nian
Li, Zhikai
Gong, Chaoyu
Zhao, Suyun
Chen, Hong
Li, Cuiping
Wang, Kai
You, Yang
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2505_06948
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Unsupervised Learning for Class Distribution Mismatch
Du, Pan
Zhao, Wangbo
Lu, Xinai
Liu, Nian
Li, Zhikai
Gong, Chaoyu
Zhao, Suyun
Chen, Hong
Li, Cuiping
Wang, Kai
You, Yang
Computer Vision and Pattern Recognition
Machine Learning
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.
title Unsupervised Learning for Class Distribution Mismatch
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2505.06948