SOrT-ing VQA Models : Contrastive Gradient Learning for Improved Consistency
This addresses a critical flaw in VQA models for AI applications requiring reliable visual understanding, though it is an incremental improvement focused on a specific domain.
The paper tackled the problem of inconsistency in Visual Question Answering (VQA) models, where they answer complex reasoning questions correctly but fail on simpler sub-questions, by proposing a contrastive gradient learning method called SOrT, which improved model consistency by up to 6.5% points over baselines.
Recent research in Visual Question Answering (VQA) has revealed state-of-the-art models to be inconsistent in their understanding of the world -- they answer seemingly difficult questions requiring reasoning correctly but get simpler associated sub-questions wrong. These sub-questions pertain to lower level visual concepts in the image that models ideally should understand to be able to answer the higher level question correctly. To address this, we first present a gradient-based interpretability approach to determine the questions most strongly correlated with the reasoning question on an image, and use this to evaluate VQA models on their ability to identify the relevant sub-questions needed to answer a reasoning question. Next, we propose a contrastive gradient learning based approach called Sub-question Oriented Tuning (SOrT) which encourages models to rank relevant sub-questions higher than irrelevant questions for an <image, reasoning-question> pair. We show that SOrT improves model consistency by upto 6.5% points over existing baselines, while also improving visual grounding.