Research map / Federated learning and privacy / Privacy attacks and secure learning
Privacy on the Fly: A Predictive Adversarial Transformation Network for Mobile Sensor Data
Tianle Song, Chenhao Lin, Yang Cao, Zhengyu Zhao, Jiahao Sun, Chong Zhang, Le Yang, Chao Shen
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 |
|---|---|---|---|
| add history-aware top-k optimization (HATO) (ablation) | PATN · MotionSense misalignment attack | ASR (%): 30.56 -> 39.43 | Improved |
| add history-aware top-k optimization (HATO) (ablation) | PATN · MotionSense misalignment attack | EER (%): 33.24 -> 40.98 | Improved |
| increase input length Tin from 10 to 30 (ablation) | PATN · MotionSense | ASR (%): 34.59 -> 40.11 | Improved |
| increase input length Tin from 30 to 50 (ablation) | PATN · MotionSense | ASR (%): 40.11 -> 30.88 | Hurt |
| use PATN instead of FGSM/PGD | PATN · MotionSense | ASR (%): 23.95 -> 40.11 | Improved |
| use PATN instead of DP | PATN · ChildShield | ASR (%): 4.12 -> 44.95 | Improved |
| use PATN instead of PrivDiffuser on benign step detection | PATN · MotionSense step detection | detected steps error: +767 -> +21 | Improved |
| transfer PATN perturbations to black-box architectures | PATN · MotionSense black-box models | ASR (%): 29.43 to 36.57 | No clear effect |
| use PGD instead of FGSM | baselines · MotionSense and ChildShield | ASR (%): 23.95 -> 23.95 | No clear effect |
Closest papers by meaning
- Towards Privacy-Preserving Visual Recognition via Adversarial Training: A Pilot Study (ECCV 2018 · similarity 0.87)
- Selective Interpretable and Motion Consistent Privacy Attribute Obfuscation for Action Recognition (CVPR 2024 · similarity 0.84)
- POLAR-Bench: A Diagnostic Benchmark for Privacy-Utility Trade-offs in LLM Agents (NeurIPS 2026 · similarity 0.84)
- MaSS: Multi-attribute Selective Suppression for Utility-preserving Data Transformation from an Information-theoretic Perspective (ICML 2024 · similarity 0.83)
- Assessing Per-Sample Membership Inference Vulnerability without Retraining (NeurIPS 2026 · similarity 0.83)
- Private Adaptive Covariance Estimation via Gaussian Graphical Models (NeurIPS 2026 · similarity 0.82)
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.
