Causal Interventional Prompt Tuning for Few-Shot Out-of-Distribution Generalization
IEEE Transactions on Pattern Analysis and Machine Intelligence · 2026-02 · 48 (2) · 1978-1991
DOI: 10.1109/tpami.2025.3621250 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026-02. Verfasst von Jie Wen und 5 weiteren Autor:innen. Die Arbeit wurde bisher 20-mal zitiert (Crossref) und nennt 80 Literaturangaben. Gefördert von: Shenzhen Science and Technology Program, Basic and Applied Basic Research Foundation of Guangdong Province, Beihang University Open Project Program of State Key Laboratory of Virtual Reality Technology and Systems.
Abstract (Originalsprache)
Fine-tuning pre-trained vision-language models (VLMs) has shown substantial benefits in a wide range of downstream tasks, often achieving impressive performance with minimal labeled data. Parameter-efficient fine-tuning techniques, in particular, have demonstrated their effectiveness in enhancing downstream task performance. However, these methods frequently struggle to generalize to out-of-distribution (OOD) data due to their reliance on non-causal representations, which can introduce biases and spurious correlations that negatively impact decision-making. Such spurious factors hinder the model's generalization ability beyond the training distribution. To address these challenges, in this paper, we propose a novel causal intervention-based prompt tuning method to adapt VLMs to few-shot OOD generalization. Specifically, we leverage the front-door adjustment technique from causal inference to mitigate the effects of spurious correlations and enhance the model's focus on causal relationships. Built upon VLMs, our approach begins by decoupling causal and non-causal representations in the vision-language alignment process. The causal representation that captures only essential semantically relevant information can serve as a mediator variable between the input image and output label, mitigating the biases from the latent confounder. To further enrich this causal representation, we propose a novel text-based diversity augmentation technique that uses textual features to provide additional semantic context. This augmentation technique can enhance the diversity of the causal representation, making it more robust and generalizable to various OOD scenarios. Experimental results across multiple OOD datasets demonstrate that our method significantly outperforms existing approaches, achieving state-of-the-art generalization performance.
Förderung
- Shenzhen Science and Technology Program
- Basic and Applied Basic Research Foundation of Guangdong Province
- Beihang University Open Project Program of State Key Laboratory of Virtual Reality Technology and Systems
- Taizhou Shenzhen Innovation Center Project
- Southwest University of Science and Technology Key Laboratory Open Project Funded
- National Natural Science Foundation of China
- Shenzhen General Research Project
Zitieren
APA
Wen, J., Liu, Y., Huang, C., Liu, C., Xu, Y., & Cao, X. (2026). Causal Interventional Prompt Tuning for Few-Shot Out-of-Distribution Generalization. IEEE Transactions on Pattern Analysis and Machine Intelligence, 48(2), 1978-1991. https://doi.org/10.1109/tpami.2025.3621250
Vancouver
Wen J, Liu Y, Huang C, Liu C, Xu Y, Cao X. Causal Interventional Prompt Tuning for Few-Shot Out-of-Distribution Generalization. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2026;48(2):1978-1991. doi:10.1109/tpami.2025.3621250
BibTeX
@article{Wen2026,
author = {Wen, Jie and Liu, Yicheng and Huang, Chao and Liu, Chengliang and Xu, Yong and Cao, Xiaochun},
title = {Causal Interventional Prompt Tuning for Few-Shot Out-of-Distribution Generalization},
journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
year = {2026},
volume = {48},
number = {2},
pages = {1978--1991},
doi = {10.1109/tpami.2025.3621250}
}
EUVIMED