Lost in Modality: Evaluating the Effectiveness of Text-Based Membership Inference Attacks on Large Multimodal Models
Large Multimodal Language Models (MLLMs) are emerging as one of the foundational tools in an expanding range of applications. Consequently, understanding training-data leakage in these systems is increasingly critical. Log-probability-based membership inference attacks (MIAs) have become a widely adopted approach for assessing data exposure in large language models (LLMs), yet their effect in MLLMs remains unclear. We present the first comprehensive evaluation of extending these text-based MIA methods to multimodal settings. Our experiments under vision-and-text (V+T) and text-only (T-only) conditions across the DeepSeek-VL and InternVL model families show that in in-distribution settings, logit-based MIAs perform comparably across configurations, with a slight V+T advantage. Conversely, in out-of-distribution settings, visual inputs act as regularizers, effectively masking membership signals.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Membership Inference Attack against Long-Context Large Language Models
Recent advances in Large Language Models (LLMs) have enabled them to overcome their context window limitations, and demonstrate exceptional retrieval and reasoning capacities on longer context. Quesion-answering systems …
Inference AttackMembership Inference AttackSemantic SimilaritySemantic Textual SimilarityCSAT‑FTCN: A Fuzzy‑Oriented Model with Contextual Self‑attention Network for Multimodal Emotion Recognition
Multimodal emotion analysis has become a hot trend because of its wide applications, such as the question-answering system. However, in a real-world scenario, people usually have mixed or partial emotions about evaluati…
Emotion RecognitionMultimodal Emotion RecognitionQuestion AnsweringAssessing Cyclostationary Malware Detection via Feature Selection and Classification
Cyclostationarity involves periodic statistical variations in signals and processes, commonly used in signal analysis and network security. In the context of attacks, cyclostationarity helps detect malicious behaviors wi…
Anomaly DetectionClassificationfeature selectionIntrusion Detection+2Image Corruption-Inspired Membership Inference Attacks against Large Vision-Language Models
Large vision-language models (LVLMs) have demonstrated outstanding performance in many downstream tasks. However, LVLMs are trained on large-scale datasets, which can pose privacy risks if training images contain sensiti…
Membership Inference for Contrastive Pre-training Models with Text-only PII Queries
Contrastive pretraining models such as CLIP and CLAP, serve as the ubiquitous perceptual backbones for modern multimodal large models, yet their reliance on web-scale data raises growing concerns about memorizing Persona…