Towards Transparent AI Systems: Interpreting Visual Question Answering Models
This addresses the need for transparency in AI systems, specifically for VQA models, but is incremental as it applies existing visualization methods to a new domain.
The paper tackled the problem of interpreting Visual Question Answering (VQA) models by identifying important input parts like pixels and words, using guided backpropagation and occlusion techniques, and found that VQA models sometimes implicitly attend to relevant regions and words without explicit mechanisms.
Deep neural networks have shown striking progress and obtained state-of-the-art results in many AI research fields in the recent years. However, it is often unsatisfying to not know why they predict what they do. In this paper, we address the problem of interpreting Visual Question Answering (VQA) models. Specifically, we are interested in finding what part of the input (pixels in images or words in questions) the VQA model focuses on while answering the question. To tackle this problem, we use two visualization techniques -- guided backpropagation and occlusion -- to find important words in the question and important regions in the image. We then present qualitative and quantitative analyses of these importance maps. We found that even without explicit attention mechanisms, VQA models may sometimes be implicitly attending to relevant regions in the image, and often to appropriate words in the question.