Diagnosing and Rectifying Vision Models using Language
This work addresses the labor-intensive process of model diagnosis in deployment settings for machine learning practitioners, offering a novel approach that is incremental in its application of existing multi-modal methods.
The paper tackles the problem of diagnosing and rectifying vision classifiers by leveraging multi-modal embedding spaces, enabling the identification of high-error data slices and influential attributes without visual data, and demonstrates effective error slice identification and classifier rectification on image datasets.
Recent multi-modal contrastive learning models have demonstrated the ability to learn an embedding space suitable for building strong vision classifiers, by leveraging the rich information in large-scale image-caption datasets. Our work highlights a distinct advantage of this multi-modal embedding space: the ability to diagnose vision classifiers through natural language. The traditional process of diagnosing model behaviors in deployment settings involves labor-intensive data acquisition and annotation. Our proposed method can discover high-error data slices, identify influential attributes and further rectify undesirable model behaviors, without requiring any visual data. Through a combination of theoretical explanation and empirical verification, we present conditions under which classifiers trained on embeddings from one modality can be equivalently applied to embeddings from another modality. On a range of image datasets with known error slices, we demonstrate that our method can effectively identify the error slices and influential attributes, and can further use language to rectify failure modes of the classifier.