Submission Type: Regular Long Paper
Submission Track: Commonsense Reasoning
Submission Track 2: Language Grounding to Vision, Robotics and Beyond
Keywords: vision language, vcr, vqa, snli-ve, visual question answering, commonsense reasoning, pretraining, multimodal, robust, low-shot, zero-shot, domain-shift, debiased, shortcut
TL;DR: Bias Mitigation in Multiple-Choice Visual Question Answering and Other Vision Language Understanding Tasks with Long Candidate Choices
Abstract: Vision-language (VL) understanding tasks evaluate models' comprehension of complex visual scenes through multiple-choice questions. However, we have identified two dataset biases that models can exploit as shortcuts to resolve various VL tasks correctly without proper understanding. The first type of dataset bias is Unbalanced Matching bias, where the correct answer overlaps the question and image more than the incorrect answers. The second type of dataset bias is Distractor Similarity bias, where incorrect answers are overly dissimilar to the correct answer but significantly similar to other incorrect answers within the same sample. To address these dataset biases, we first propose Adversarial Data Synthesis (ADS) to generate synthetic training and debiased evaluation data. We then introduce Intra-sample Counterfactual Training (ICT) to assist models in utilizing the synthesized training data, particularly the counterfactual data, via focusing on intra-sample differentiation. Extensive experiments demonstrate the effectiveness of ADS and ICT in consistently improving model performance across different benchmarks, even in domain-shifted scenarios.
Submission Number: 4198
Loading