Diffexplainer: Towards Cross-modal Global Explanations with Diffusion Models

arXiv:2404.02618v1 Announce Type: new
Abstract: We present DiffExplainer, a novel framework that, leveraging language-vision models, enables multimodal global explainability. DiffExplainer employs diffusion models conditioned on optimized text prompts, synthesizing images that maximize class outputs and hidden features of a classifier, thus providing a visual tool for explaining decisions. Moreover, the analysis of generated visual descriptions allows for automatic identification of biases and spurious features, as opposed to traditional methods that often rely on manual intervention. The cross-modal transferability of language-vision …

abstract analysis arxiv class classifier cs.ai cs.cv decisions diffusion diffusion models explainability features framework global hidden images language modal multimodal novel prompts text tool type vision vision models visual