Towards More Faithful Natural Language Explanation Using Multi-Level Contrastive Learning in VQA

TL;DR

This paper introduces MCLE, a multi-level contrastive learning approach that improves the faithfulness of natural language explanations in VQA by aligning explanations more closely with visual and factual data.

Contribution

The paper proposes a novel self-supervised contrastive learning framework that enhances explanation faithfulness in VQA by leveraging multi-level semantic, image, and instance-level samples.

Findings

Improves explanation faithfulness and logical consistency.

Achieves better alignment between explanations and visual facts.

Outperforms existing methods on VQA-NLE benchmarks.

Abstract

Natural language explanation in visual question answer (VQA-NLE) aims to explain the decision-making process of models by generating natural language sentences to increase users' trust in the black-box systems. Existing post-hoc methods have achieved significant progress in obtaining a plausible explanation. However, such post-hoc explanations are not always aligned with human logical inference, suffering from the issues on: 1) Deductive unsatisfiability, the generated explanations do not logically lead to the answer; 2) Factual inconsistency, the model falsifies its counterfactual explanation for answers without considering the facts in images; and 3) Semantic perturbation insensitivity, the model can not recognize the semantic changes caused by small perturbations. These problems reduce the faithfulness of explanations generated by models. To address the above issues, we propose a novel self-supervised \textbf{M}ulti-level \textbf{C}ontrastive \textbf{L}earning based natural language \textbf{E}xplanation model (MCLE) for VQA with semantic-level, image-level, and instance-level factual and counterfactual samples. MCLE extracts discriminative features and aligns the feature spaces from explanations with visual question and answer to generate more consistent explanations. We conduct extensive experiments, ablation analysis, and case study to demonstrate the effectiveness of our method on two VQA-NLE benchmarks.