PDA: Text-Augmented Defense Framework for Robust Vision-Language Models against Adversarial Image Attacks

TL;DR

PDA is a training-free, text-augmented defense framework that enhances the robustness of vision-language models against adversarial image attacks during inference.

Contribution

Introducing PDA, a novel inference-time defense method using text augmentation to improve VLM robustness without model modification.

Findings

PDA improves robustness against various adversarial attacks across multiple VLM architectures.

PDA maintains high accuracy on clean images while defending against adversarial perturbations.

PDA is computationally efficient and applicable during inference without retraining.

Abstract

Vision-language models (VLMs) are vulnerable to adversarial image perturbations. Existing works based on adversarial training against task-specific adversarial examples are computationally expensive and often fail to generalize to unseen attack types. To address these limitations, we introduce Paraphrase-Decomposition-Aggregation (PDA), a training-free defense framework that leverages text augmentation to enhance VLM robustness under diverse adversarial image attacks. PDA performs prompt paraphrasing, question decomposition, and consistency aggregation entirely at test time, thus requiring no modification on the underlying models. To balance robustness and efficiency, we instantiate PDA as invariants that reduce the inference cost while retaining most of its robustness gains. Experiments on multiple VLM architectures and benchmarks for visual question answering, classification, and captioning show that PDA achieves consistent robustness gains against various adversarial perturbations while maintaining competitive clean accuracy, establishing a generic, strong and practical defense framework for VLMs during inference.