Multimodal Explanations: Justifying Decisions and Pointing to the Evidence
This work addresses the need for explainable AI in multimodal settings, offering a novel approach that could benefit applications like activity recognition and visual question answering, though it is incremental in building on prior unimodal methods.
The authors tackled the problem of making deep models both effective and explainable by proposing a multimodal approach that combines textual rationale generation and attention visualization, showing that training with textual explanations improves both justification models and evidence localization on new datasets for activity recognition and visual question answering tasks.
Deep models that are both effective and explainable are desirable in many settings; prior explainable models have been unimodal, offering either image-based visualization of attention weights or text-based generation of post-hoc justifications. We propose a multimodal approach to explanation, and argue that the two modalities provide complementary explanatory strengths. We collect two new datasets to define and evaluate this task, and propose a novel model which can provide joint textual rationale generation and attention visualization. Our datasets define visual and textual justifications of a classification decision for activity recognition tasks (ACT-X) and for visual question answering tasks (VQA-X). We quantitatively show that training with the textual explanations not only yields better textual justification models, but also better localizes the evidence that supports the decision. We also qualitatively show cases where visual explanation is more insightful than textual explanation, and vice versa, supporting our thesis that multimodal explanation models offer significant benefits over unimodal approaches.