article · Scientific Reports
Using cloud-based machine learning services often exposes sensitive user data to security risks because servers process data in plaintext. HE-CloudML is a secure framework designed to carry out deep neural network inference directly on encrypted cloud data using homomorphic encryption. The system uses a three-tier architecture that combines client-side encryption, an encrypted cloud inference engine, and distributed key management to keep raw data hidden from the server throughout the process. To handle complex calculations efficiently, the framework incorporates polynomial activation functions, optimised ciphertext batching, and an adaptive lazy bootstrapping pipeline that cuts latency and computational depth. Tested across image classification, medical imaging, and financial fraud detection, the framework achieves accuracy comparable to unencrypted models, showing a maximum drop in accuracy of 1.81 percent while running up to 26.9 times faster than the CryptoNets baseline.
Organisations handling confidential information, such as medical records or financial transactions, often cannot safely use cloud machine learning without exposing private data. This framework demonstrates that advanced encryption can protect sensitive user inputs during cloud analysis without sacrificing high accuracy or incurring prohibitive operational delays, helping to make secure cloud computing practical.
This work is an applied and tested framework with direct relevance to cloud service providers, healthcare systems, and financial institutions handling sensitive datasets. It could enable privacy-compliant Machine Learning as a Service for fraud detection and diagnostic imaging. While the framework has been successfully validated against multiple domain-specific benchmarks, further engineering would be required to integrate it into production-grade cloud environments.
AI-generated from the published abstract. Always read the original work before citing.
Abstract The widespread adoption of cloud-based Machine Learning as a Service (MLaaS) exposes sensitive user data to critical privacy risks during inference, as plaintext data must typically be processed by untrusted cloud servers. This paper presents HE-CloudML, a unified privacy-preserving framework for secure deep neural network (DNN) inference over encrypted cloud data using Homomorphic Encryption (HE). HE-CloudML is architected as a three-tier system comprising a client-side CKKS encryption module, a cloud-side HE inference engine, and a distributed key management layer, ensuring that raw input data is never exposed to the server at any stage of computation. The framework introduces HE-compatible polynomial activation function approximations via degree-5 Chebyshev minimax polynomials, an optimized SIMD ciphertext batching strategy exploiting Ring Learning With Errors (RLWE) slot packing, and an adaptive lazy bootstrapping pipeline to substantially reduce homomorphic evaluation depth and inference latency. A formal security analysis under the IND-CPA model grounded in the RLWE hardness assumption demonstrates resistance to inference, model inversion, and membership inference attacks. Comprehensive experiments across three domains benchmark image classification (MNIST: 99.28%, CIFAR-10: 90.37%), medical imaging (93.61%), and financial fraud detection (96.44%) demonstrate that HE-CloudML achieves near-plaintext accuracy with a maximum accuracy drop of 1.81%, while delivering up to 26.9× latency improvements over CryptoNets.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.1038/s41598-026-69466-7
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
