How Transferable are Reasoning Patterns in VQA?

TL;DR

This paper investigates the transferability of reasoning patterns in VQA, demonstrating that transferring insights from a less bias-prone visual oracle improves model accuracy and reduces bias reliance.

Contribution

It introduces a visual oracle for studying reasoning in VQA, analyzes attention mechanisms, and shows that transferring reasoning patterns enhances model performance and generalization.

Findings

Transferred reasoning patterns improve overall accuracy.

Model shows increased accuracy on infrequent answers.

Reduced dependency on dataset biases.

Abstract

Since its inception, Visual Question Answering (VQA) is notoriously known as a task, where models are prone to exploit biases in datasets to find shortcuts instead of performing high-level reasoning. Classical methods address this by removing biases from training data, or adding branches to models to detect and remove biases. In this paper, we argue that uncertainty in vision is a dominating factor preventing the successful learning of reasoning in vision and language problems. We train a visual oracle and in a large scale study provide experimental evidence that it is much less prone to exploiting spurious dataset biases compared to standard models. We propose to study the attention mechanisms at work in the visual oracle and compare them with a SOTA Transformer-based model. We provide an in-depth analysis and visualizations of reasoning patterns obtained with an online visualization tool which we make publicly available (https://reasoningpatterns.github.io). We exploit these insights by transferring reasoning patterns from the oracle to a SOTA Transformer-based VQA model taking standard noisy visual inputs via fine-tuning. In experiments we report higher overall accuracy, as well as accuracy on infrequent answers for each question type, which provides evidence for improved generalization and a decrease of the dependency on dataset biases.