English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Fully homomorphic encryption (FHE) is a promising cryptographic primitive forrealizing private neural network inference (PI) services by allowing a clientto fully offload the inference task to a cloud server while keeping the clientdata oblivious to the server. This work proposes NeuJeans, an FHE-basedsolution for the PI of deep convolutional neural networks (CNNs). NeuJeanstackles the critical problem of the enormous computational cost for the FHEevaluation of convolutional layers (conv2d), mainly due to the high cost ofdata reordering and bootstrapping. We first propose an encoding methodintroducing nested structures inside encoded vectors for FHE, which enables usto develop efficient conv2d algorithms with reduced data reordering costs.However, the new encoding method also introduces additional computations forconversion between encoding methods, which could negate its advantages. Wediscover that fusing conv2d with bootstrapping eliminates such computationswhile reducing the cost of bootstrapping. Then, we devise optimized executionflows for various types of conv2d and apply them to end-to-end implementationof CNNs. NeuJeans accelerates the performance of conv2d by up to 5.68 timescompared to state-of-the-art FHE-based PI work and performs the PI of a CNN atthe scale of ImageNet (ResNet18) within a mere few seconds

NeuJeans: Private Neural Network Inference with Joint Optimization of Convolution and Bootstrapping