Good, Better, Best: Textual Distractors Generation for Multiple-Choice Visual Question Answering via Reinforcement Learning
This addresses the need for automatically constructing large-scale multiple-choice VQA data, which is important for researchers and end-users in AI applications like open-domain scenarios and security-sensitive tasks, but it is incremental as it builds on existing VQA models and reinforcement learning techniques.
The paper tackles the problem of generating challenging textual distractors for multiple-choice visual question answering (VQA) without ground-truth training data, proposing a reinforcement learning framework called Gobbet that uses pre-trained VQA models as feedback, and it shows utility through data augmentation experiments and a manual case study.
Multiple-choice VQA has drawn increasing attention from researchers and end-users recently. As the demand for automatically constructing large-scale multiple-choice VQA data grows, we introduce a novel task called textual Distractors Generation for VQA (DG-VQA) focusing on generating challenging yet meaningful distractors given the context image, question, and correct answer. The DG-VQA task aims at generating distractors without ground-truth training samples since such resources are rarely available. To tackle the DG-VQA unsupervisedly, we propose Gobbet, a reinforcement learning(RL) based framework that utilizes pre-trained VQA models as an alternative knowledge base to guide the distractor generation process. In Gobbet, a pre-trained VQA model serves as the environment in RL setting to provide feedback for the input multi-modal query, while a neural distractor generator serves as the agent to take actions accordingly. We propose to use existing VQA models' performance degradation as indicators of the quality of generated distractors. On the other hand, we show the utility of generated distractors through data augmentation experiments, since robustness is more and more important when AI models apply to unpredictable open-domain scenarios or security-sensitive applications. We further conduct a manual case study on the factors why distractors generated by Gobbet can fool existing models.