RUBi: Reducing Unimodal Biases in Visual Question Answering
Authors:
Preprint
Abstract
Visual Question Answering (VQA) is the task of answering questions about an image. Some VQA models often exploit unimodal biases to provide the correct answer without using the image information. As a result, they suffer from a huge drop in performance when evaluated on data outside their training set distribution. This critical issue makes them unsuitable for real-world settings. We propose RUBi, a new learning strategy to reduce biases in any VQA model. It reduces the importance of the most biased examples, i.e. examples that can be correctly classified without looking at the image. It implicitly forces the VQA model to use the two input modalities instead of relying on statistical regularities between the question and the answer. We leverage a question-only model that captures the language biases by identifying when these unwanted regularities are used. It prevents the base VQA model from learning them by influencing its predictions. This leads to dynamically adjusting the loss in order to compensate for biases. We validate our contributions by surpassing the current state-of-the-art results on VQA-CP v2. This dataset is specifically designed to assess the robustness of VQA models when exposed to different question biases at test time than what was seen during training. Our code is available: github.com/cdancette/rubi.bootstrap.pytorchKeywords
All Related Versions
This publication has 10 references indexed in Scilit:
- Explainable neural computation via stack neural module networksApplied AI Letters, 2021
- From Recognition to Cognition: Visual Commonsense Reasoning2019
- Answer Them All! Toward Universal Visual Question Answering Models2019
- Explainable and Explicit Visual Reasoning Over Scene Graphs2019
- Explicit Bias Discovery in Visual Question Answering Models2019
- Beyond Bilinear: Generalized Multimodal Factorized High-Order Pooling for Visual Question AnsweringIEEE Transactions on Neural Networks and Learning Systems, 2018
- Being Negative but Constructively: Lessons Learnt from Creating Better Visual Question Answering Datasets2018
- MUTAN: Multimodal Tucker Fusion for Visual Question Answering2017
- Analyzing the Behavior of Visual Question Answering Models2016
- Unbiased look at dataset bias2011