Neuro-Symbolic Visual Reasoning: Disentangling "Visual" from "Reasoning"
This addresses the challenge of improving and assessing reasoning independently in VQA, which is incremental as it builds on neuro-symbolic models like Neural Module Networks.
The paper tackled the problem of disentangling visual perception from reasoning in visual question answering (VQA) by proposing a framework to isolate and evaluate reasoning separately, and a top-down calibration technique to handle imperfect perception, leading to in-depth comparisons on the GQA dataset.
Visual reasoning tasks such as visual question answering (VQA) require an interplay of visual perception with reasoning about the question semantics grounded in perception. However, recent advances in this area are still primarily driven by perception improvements (e.g. scene graph generation) rather than reasoning. Neuro-symbolic models such as Neural Module Networks bring the benefits of compositional reasoning to VQA, but they are still entangled with visual representation learning, and thus neural reasoning is hard to improve and assess on its own. To address this, we propose (1) a framework to isolate and evaluate the reasoning aspect of VQA separately from its perception, and (2) a novel top-down calibration technique that allows the model to answer reasoning questions even with imperfect perception. To this end, we introduce a differentiable first-order logic formalism for VQA that explicitly decouples question answering from visual perception. On the challenging GQA dataset, this framework is used to perform in-depth, disentangled comparisons between well-known VQA models leading to informative insights regarding the participating models as well as the task.