Saved in:
Bibliographic Details
Main Authors: Wang, Yinong Oliver, Li, Eileen, Luo, Jinqi, Wang, Zhaoning, De la Torre, Fernando
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2410.06243
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916428542640128
author Wang, Yinong Oliver
Li, Eileen
Luo, Jinqi
Wang, Zhaoning
De la Torre, Fernando
author_facet Wang, Yinong Oliver
Li, Eileen
Luo, Jinqi
Wang, Zhaoning
De la Torre, Fernando
contents Ensuring model explainability and robustness is essential for reliable deployment of deep vision systems. Current methods for evaluating robustness rely on collecting and annotating extensive test sets. While this is common practice, the process is labor-intensive and expensive with no guarantee of sufficient coverage across attributes of interest. Recently, model diagnosis frameworks have emerged leveraging user inputs (e.g., text) to assess the vulnerability of the model. However, such dependence on human can introduce bias and limitation given the domain knowledge of particular users. This paper proposes Unsupervised Model Diagnosis (UMO), that leverages generative models to produce semantic counterfactual explanations without any user guidance. Given a differentiable computer vision model (i.e., the target model), UMO optimizes for the most counterfactual directions in a generative latent space. Our approach identifies and visualizes changes in semantics, and then matches these changes to attributes from wide-ranging text sources, such as dictionaries or language models. We validate the framework on multiple vision tasks (e.g., classification, segmentation, keypoint detection). Extensive experiments show that our unsupervised discovery of semantic directions can correctly highlight spurious correlations and visualize the failure mode of target models without any human intervention.
format Preprint
id arxiv_https___arxiv_org_abs_2410_06243
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Unsupervised Model Diagnosis
Wang, Yinong Oliver
Li, Eileen
Luo, Jinqi
Wang, Zhaoning
De la Torre, Fernando
Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
Machine Learning
Ensuring model explainability and robustness is essential for reliable deployment of deep vision systems. Current methods for evaluating robustness rely on collecting and annotating extensive test sets. While this is common practice, the process is labor-intensive and expensive with no guarantee of sufficient coverage across attributes of interest. Recently, model diagnosis frameworks have emerged leveraging user inputs (e.g., text) to assess the vulnerability of the model. However, such dependence on human can introduce bias and limitation given the domain knowledge of particular users. This paper proposes Unsupervised Model Diagnosis (UMO), that leverages generative models to produce semantic counterfactual explanations without any user guidance. Given a differentiable computer vision model (i.e., the target model), UMO optimizes for the most counterfactual directions in a generative latent space. Our approach identifies and visualizes changes in semantics, and then matches these changes to attributes from wide-ranging text sources, such as dictionaries or language models. We validate the framework on multiple vision tasks (e.g., classification, segmentation, keypoint detection). Extensive experiments show that our unsupervised discovery of semantic directions can correctly highlight spurious correlations and visualize the failure mode of target models without any human intervention.
title Unsupervised Model Diagnosis
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
Machine Learning
url https://arxiv.org/abs/2410.06243