Research map / Federated learning and privacy / Privacy attacks and secure learning
CryptoNets: Applying Neural Networks to Encrypted Data with High Throughput and Accuracy
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| batch 4096 instances via SIMD (CRT polynomial packing) | CryptoNets · MNIST encrypted inference | throughput: 58982 per hour | Helped |
| replace sigmoid output with nothing at prediction time | CryptoNets · MNIST | prediction correctness | No clear effect |
| compose linear layers 3 to 6 into a single matrix | CryptoNets · MNIST | network depth: 9 layers -> 5 layers | Helped |
| add convolution layers without activation (layers 4 and 5) | square-activation network · MNIST | overfitting | Helped |
| use square activation instead of ReLU/sigmoid | CryptoNets · MNIST | trainability | Mixed |
| encrypt data with homomorphic encryption | CryptoNets · MNIST | accuracy: 99% | No clear effect |
| batch 4096 images in one encrypted message | CryptoNets · MNIST | per-image message size: 367.5 MB -> 91.875 KB per image | Helped |
| encrypt rather than use raw double-precision images | CryptoNets · MNIST | message size: 15.3 times larger | Hurt |
Closest papers by meaning
- DCT-CryptoNets: Scaling Private Inference in the Frequency Domain (ICLR 2025 · similarity 0.84)
- United We Stand: Accelerating Privacy-Preserving Neural Inference by Conjunctive Optimization with Interleaved Nexus (AAAI 2024 · similarity 0.81)
- HEPrune: Fast Private Training of Deep Neural Networks With Encrypted Data Pruning (NeurIPS 2024 · similarity 0.81)
- Train to Defend: First Defense Against Cryptanalytic Neural Network Parameter Extraction Attacks (NeurIPS 2025 · similarity 0.81)
- FicGCN: Unveiling the Homomorphic Encryption Efficiency from Irregular Graph Convolutional Networks (ICML 2025 · similarity 0.81)
- Converting Transformers to Polynomial Form for Secure Inference Over Homomorphic Encryption (ICML 2024 · similarity 0.80)
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.