Rethinking Misalignment in Vision-Language Model Adaptation from a Causal Perspective

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Yanan, Li, Jiangmeng, Liu, Lixiang, Qiang, Wenwen
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913571616587776
author Zhang, Yanan
Li, Jiangmeng
Liu, Lixiang
Qiang, Wenwen
author_facet Zhang, Yanan
Li, Jiangmeng
Liu, Lixiang
Qiang, Wenwen
contents Foundational Vision-Language models such as CLIP have exhibited impressive generalization in downstream tasks. However, CLIP suffers from a two-level misalignment issue, i.e., task misalignment and data misalignment, when adapting to specific tasks. Soft prompt tuning has mitigated the task misalignment, yet the data misalignment remains a challenge. To analyze the impacts of the data misalignment, we revisit the pre-training and adaptation processes of CLIP and develop a structural causal model. We discover that while we expect to capture task-relevant information for downstream tasks accurately, the task-irrelevant knowledge impacts the prediction results and hampers the modeling of the true relationships between the images and the predicted classes. As task-irrelevant knowledge is unobservable, we leverage the front-door adjustment and propose Causality-Guided Semantic Decoupling and Classification (CDC) to mitigate the interference of task-irrelevant knowledge. Specifically, we decouple semantics contained in the data of downstream tasks and perform classification based on each semantic. Furthermore, we employ the Dempster-Shafer evidence theory to evaluate the uncertainty of each prediction generated by diverse semantics. Experiments conducted in multiple different settings have consistently demonstrated the effectiveness of CDC.
format Preprint
id arxiv_https___arxiv_org_abs_2410_12816
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Rethinking Misalignment in Vision-Language Model Adaptation from a Causal Perspective
Zhang, Yanan
Li, Jiangmeng
Liu, Lixiang
Qiang, Wenwen
Computer Vision and Pattern Recognition
Computation and Language
Machine Learning
Foundational Vision-Language models such as CLIP have exhibited impressive generalization in downstream tasks. However, CLIP suffers from a two-level misalignment issue, i.e., task misalignment and data misalignment, when adapting to specific tasks. Soft prompt tuning has mitigated the task misalignment, yet the data misalignment remains a challenge. To analyze the impacts of the data misalignment, we revisit the pre-training and adaptation processes of CLIP and develop a structural causal model. We discover that while we expect to capture task-relevant information for downstream tasks accurately, the task-irrelevant knowledge impacts the prediction results and hampers the modeling of the true relationships between the images and the predicted classes. As task-irrelevant knowledge is unobservable, we leverage the front-door adjustment and propose Causality-Guided Semantic Decoupling and Classification (CDC) to mitigate the interference of task-irrelevant knowledge. Specifically, we decouple semantics contained in the data of downstream tasks and perform classification based on each semantic. Furthermore, we employ the Dempster-Shafer evidence theory to evaluate the uncertainty of each prediction generated by diverse semantics. Experiments conducted in multiple different settings have consistently demonstrated the effectiveness of CDC.
title Rethinking Misalignment in Vision-Language Model Adaptation from a Causal Perspective
topic Computer Vision and Pattern Recognition
Computation and Language
Machine Learning
url https://arxiv.org/abs/2410.12816