Faster CryptoNets: Leveraging Sparsity for Real-World Encrypted Inference
read the original abstract
Homomorphic encryption enables arbitrary computation over data while it remains encrypted. This privacy-preserving feature is attractive for machine learning, but requires significant computational time due to the large overhead of the encryption scheme. We present Faster CryptoNets, a method for efficient encrypted inference using neural networks. We develop a pruning and quantization approach that leverages sparse representations in the underlying cryptosystem to accelerate inference. We derive an optimal approximation for popular activation functions that achieves maximally-sparse encodings and minimizes approximation error. We also show how privacy-safe training techniques can be used to reduce the overhead of encrypted inference for real-world datasets by leveraging transfer learning and differential privacy. Our experiments show that our method maintains competitive accuracy and achieves a significant speedup over previous methods. This work increases the viability of deep learning systems that use homomorphic encryption to protect user privacy.
This paper has not been read by Pith yet.
Forward citations
Cited by 4 Pith papers
-
Volley Revolver: A Novel Matrix-Encoding Method for Privacy-Preserving Neural Networks (Inference)
A matrix-encoding method for homomorphic encryption enables privacy-preserving CNN inference, with ~287s runtime reported for classifying 32 encrypted 28x28 MNIST images on a 40-vCPU server.
-
Towards Characterizing and Limiting Information Exposure in DNN Layers
Framework quantifies per-layer sensitive information exposure in DNNs via generalization error and evaluates TEE-based protection for the most exposed layers against white-box membership inference.
-
Kernel-Based ReLU Approximation for Homomorphic Encryption-Compatible Privacy-preserving Deep Learning Models
Kernel-based ReLU is approximated by a quadratic polynomial for low-depth homomorphic encryption compatibility, trained on LLM token embeddings and evaluated across DL and transformer settings.
-
Towards Deep Encrypted Training: Low-Latency, Memory-Efficient, and High-Throughput Inference for Privacy-Preserving Neural Networks
Optimized batched homomorphic encryption with a new pipeline architecture delivers 1.78x faster amortized inference and 3.74x lower memory than prior work for ResNet-20 on 512 encrypted CIFAR-10 images.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.