Research map / Federated learning and privacy / Privacy attacks and secure learning
$\texttt{FHAIM}$: Fully Homomorphic AIM for Private Tabular Synthetic Data Generation
Mayank Kumar, Qian Lou, Paulo L. Barreto, Martine De Cock, Sikha Pentyala
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 |
|---|---|---|---|
| use exact squared L2 norm instead of polynomial-approximated L1 norm (ablation) | FHAIM · breast-cancer, COMPAS, diabetes (eps = infinity) | workload error / classification accuracy | Improved |
| use exact squared L2 norm instead of polynomial-approximated L1 norm (ablation) | FHAIM · eps = 1 (high privacy) | classification accuracy | Improved |
| switch from L1 to L2 norm in SELECT step (ablation) | FHAIM · Cancer | runtime per iteration: about 15% increase | Hurt |
| switch from L1 to L2 norm (ablation) | FHAIM · COMPAS | peak memory (MB): 11.12 -> 31.72 | Hurt |
| run FHAIM encrypted instead of plaintext AIM-L1 | AIM · Adult | generation time: 19 minutes -> 5.5 hours | Hurt |
| run FHAIM encrypted instead of plaintext AIM-L2 | AIM · Adult | generation time: 5 minutes -> 5.3 hours | Hurt |
| add encryption of the data via FHE (FHAIM-L2 vs AIM-L2) | AIM · three datasets | utility | No clear effect |
Closest papers by meaning
- CaPS: Collaborative and Private Synthetic Data Generation from Distributed Sources (ICML 2024 · similarity 0.88)
- Encryption-Friendly LLM Architecture (ICLR 2025 · similarity 0.87)
- EncryptedLLM: Privacy-Preserving Large Language Model Inference via GPU-Accelerated Fully Homomorphic Encryption (ICML 2025 · similarity 0.87)
- HARMONIC: Harnessing LLMs for Tabular Data Synthesis and Privacy Protection (NeurIPS 2024 · similarity 0.87)
- A Privacy-Preserving Framework for Generative Model-driven Synthetic Datasets (AAAI 2025 · similarity 0.86)
- Converting Transformers to Polynomial Form for Secure Inference Over Homomorphic Encryption (ICML 2024 · similarity 0.86)
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.
