Research map / Federated learning and privacy / Privacy attacks and secure learning
The Hidden Risk: Membership Inference Attacks on Multimodal Federated Learning via Modality Imbalance
chang Ma, Jun Li, Kang Wei, Yipeng Zhou, Ming Ding, Yiyang Ni
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| attack only the dominant (audio) modality instead of the whole multimodal model (ablation) | Cosine-LIRA · CREMA-D | AUC: 0.7500 -> 0.7551 | No clear effect |
| attack only the non-dominant (visual) modality (ablation) | Cosine-LIRA · CREMA-D | AUC: 0.7500 -> 0.5140 | Hurt |
| add modality-aware adaptive weighting and calibration of MIA scores | MIA on multimodal FL · CREMA-D 6 clients | AUC: 0.7500 -> 0.7802 | Improved |
| add modality-aware attack model | MIA on multimodal FL · CREMA-D 6 clients | TPR @ 1% FPR: 12.28 -> 16.67 | Improved |
| reduce number of colluding clients from 10 to 2 (ablation) | proposed MIA · Balanced dataset | TPR @ 1% FPR: 14.86 -> 11.86 (ASL-Gating) | Hurt |
| reduce reference models from 6 to 0 (ablation) | Likelihood-based MIA (Cosine-LiRA M) · CREMA-D-10 | TPR @ 1% FPR: 10.01 -> 0.00 | Hurt |
| reduce reference models from 6 to 0 (ablation) | proposed MIA · CREMA-D-10 | TPR @ 1% FPR: 16.33 -> 15.32 | No clear effect |
| more severe overfitting (ablation) | proposed MIA · CREMA-D-10 | AUC: .6473 -> .8576 | Improved |
| non-IID modality-quality heterogeneity across clients (ablation) | MIAs generally · CREMA-D-6 | AUC: Cosine-LiRA (M) client 0: .8769 | Improved |
| add OGM modality balancing as defense (ablation) | MFL · CREMA-D | attack AUC: 0.7798 -> 0.5697 | Improved |
| add uni-modal loss regularization as defense (ablation) | MFL · CREMA-D | model accuracy: 55.17 -> 56.11 | Improved |
| add client-level DP with sigma 0.1 (ablation) | MFL · CREMA-D | model accuracy: 55.17 -> 34.63 | Hurt |
Closest papers by meaning
- Adaptive Hyper-graph Aggregation for Modality-Agnostic Federated Learning (CVPR 2024 · similarity 0.86)
- FedMBridge: Bridgeable Multimodal Federated Learning (ICML 2024 · similarity 0.86)
- Incongruent Multimodal Federated Learning for Medical Vision and Language-based Multi-label Disease Detection (AAAI 2025 · similarity 0.85)
- FedMIA: An Effective Membership Inference Attack Exploiting "All for One" Principle in Federated Learning (CVPR 2025 · similarity 0.85)
- Enhancing Storage and Computational Efficiency in Federated Multimodal Learning for Large-Scale Models (ICML 2024 · similarity 0.85)
- Learning Reconfigurable Representations for Multimodal Federated Learning with Missing Data (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
