CVAug 7, 2023

TIJO: Trigger Inversion with Joint Optimization for Defending Multimodal Backdoored Models

arXiv:2308.03906v117 citationsh-index: 37Has Code
Originality Incremental advance
AI Analysis

This addresses security vulnerabilities in multimodal AI systems, particularly for visual question answering, with incremental improvements over existing unimodal defenses.

The paper tackles the problem of defending multimodal models against dual-key backdoor attacks by proposing TIJO, a joint optimization method that reverse-engineers triggers in both image and text modalities, improving AUC from 0.6 to 0.92 on the TrojVQA benchmark.

We present a Multimodal Backdoor Defense technique TIJO (Trigger Inversion using Joint Optimization). Recent work arXiv:2112.07668 has demonstrated successful backdoor attacks on multimodal models for the Visual Question Answering task. Their dual-key backdoor trigger is split across two modalities (image and text), such that the backdoor is activated if and only if the trigger is present in both modalities. We propose TIJO that defends against dual-key attacks through a joint optimization that reverse-engineers the trigger in both the image and text modalities. This joint optimization is challenging in multimodal models due to the disconnected nature of the visual pipeline which consists of an offline feature extractor, whose output is then fused with the text using a fusion module. The key insight enabling the joint optimization in TIJO is that the trigger inversion needs to be carried out in the object detection box feature space as opposed to the pixel space. We demonstrate the effectiveness of our method on the TrojVQA benchmark, where TIJO improves upon the state-of-the-art unimodal methods from an AUC of 0.6 to 0.92 on multimodal dual-key backdoors. Furthermore, our method also improves upon the unimodal baselines on unimodal backdoors. We present ablation studies and qualitative results to provide insights into our algorithm such as the critical importance of overlaying the inverted feature triggers on all visual features during trigger inversion. The prototype implementation of TIJO is available at https://github.com/SRI-CSL/TIJO.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes