Text-Driven Causal Representation Learning for Source-Free Domain Generalization

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Zhou, Lihua, Ye, Mao, Li, Nianxin, Li, Shuaifeng, Wu, Jinlin, Zhu, Xiatian, Deng, Lei, Liu, Hongbin, Luo, Jiebo, Lei, Zhen
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912481127956480
author Zhou, Lihua
Ye, Mao
Li, Nianxin
Li, Shuaifeng
Wu, Jinlin
Zhu, Xiatian
Deng, Lei
Liu, Hongbin
Luo, Jiebo
Lei, Zhen
author_facet Zhou, Lihua
Ye, Mao
Li, Nianxin
Li, Shuaifeng
Wu, Jinlin
Zhu, Xiatian
Deng, Lei
Liu, Hongbin
Luo, Jiebo
Lei, Zhen
contents Deep learning often struggles when training and test data distributions differ. Traditional domain generalization (DG) tackles this by including data from multiple source domains, which is impractical due to expensive data collection and annotation. Recent vision-language models like CLIP enable source-free domain generalization (SFDG) by using text prompts to simulate visual representations, reducing data demands. However, existing SFDG methods struggle with domain-specific confounders, limiting their generalization capabilities. To address this issue, we propose TDCRL (\textbf{T}ext-\textbf{D}riven \textbf{C}ausal \textbf{R}epresentation \textbf{L}earning), the first method to integrate causal inference into the SFDG setting. TDCRL operates in two steps: first, it employs data augmentation to generate style word vectors, combining them with class information to generate text embeddings to simulate visual representations; second, it trains a causal intervention network with a confounder dictionary to extract domain-invariant features. Grounded in causal learning, our approach offers a clear and effective mechanism to achieve robust, domain-invariant features, ensuring robust generalization. Extensive experiments on PACS, VLCS, OfficeHome, and DomainNet show state-of-the-art performance, proving TDCRL effectiveness in SFDG.
format Preprint
id arxiv_https___arxiv_org_abs_2507_09961
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Text-Driven Causal Representation Learning for Source-Free Domain Generalization
Zhou, Lihua
Ye, Mao
Li, Nianxin
Li, Shuaifeng
Wu, Jinlin
Zhu, Xiatian
Deng, Lei
Liu, Hongbin
Luo, Jiebo
Lei, Zhen
Machine Learning
Deep learning often struggles when training and test data distributions differ. Traditional domain generalization (DG) tackles this by including data from multiple source domains, which is impractical due to expensive data collection and annotation. Recent vision-language models like CLIP enable source-free domain generalization (SFDG) by using text prompts to simulate visual representations, reducing data demands. However, existing SFDG methods struggle with domain-specific confounders, limiting their generalization capabilities. To address this issue, we propose TDCRL (\textbf{T}ext-\textbf{D}riven \textbf{C}ausal \textbf{R}epresentation \textbf{L}earning), the first method to integrate causal inference into the SFDG setting. TDCRL operates in two steps: first, it employs data augmentation to generate style word vectors, combining them with class information to generate text embeddings to simulate visual representations; second, it trains a causal intervention network with a confounder dictionary to extract domain-invariant features. Grounded in causal learning, our approach offers a clear and effective mechanism to achieve robust, domain-invariant features, ensuring robust generalization. Extensive experiments on PACS, VLCS, OfficeHome, and DomainNet show state-of-the-art performance, proving TDCRL effectiveness in SFDG.
title Text-Driven Causal Representation Learning for Source-Free Domain Generalization
topic Machine Learning
url https://arxiv.org/abs/2507.09961