PrivDNN: A Secure Multi-Party Computation Framework for Deep Learning using Partial DNN Encryption

2026-07-23

AI Summary

This work addresses the challenge of inefficient privacy-preserving inference in scenarios where both deep learning models and user data must remain confidential. Existing approaches often incur substantial computational overhead and low efficiency. To overcome this, the authors propose an efficient privacy-preserving inference mechanism that identifies critical neurons and selectively encrypts only the most relevant model components. By integrating secure multi-party computation with homomorphic encryption, the method offloads ciphertext-domain computations to the user side. This strategy significantly accelerates inference while preserving model accuracy and privacy guarantees, achieving up to a 97% reduction in both inference latency and memory consumption.

Abstract

In the past decade, we have witnessed an exponential growth of deep learning models, platforms, and applications. While existing DL applications and Machine Learning as a service (MLaaS) frameworks assume fully trusted models, the need for privacy-preserving DNN evaluation arises. In a secure multi-party computation scenario, both the model and the data are considered proprietary, i.e., the model owner does not want to reveal the highly valuable DL model to the user, while the user does not wish to disclose their private data samples either. Conventional privacy-preserving deep learning solutions ask the users to send encrypted samples to the model owners, who must handle the heavy lifting of ciphertext-domain computation with homomorphic encryption. In this paper, we present a novel solution, namely, PrivDNN, which (1) offloads the computation to the user side by sharing an encrypted deep learning model with them, (2) significantly improves the efficiency of DNN evaluation using partial DNN encryption, (3) ensures model accuracy and model privacy using a core neuron selection and encryption scheme. Experimental results show that PrivDNN reduces privacy-preserving DNN inference time and memory requirement by up to 97% while maintaining model performance and privacy. Codes can be found at https://github.com/LiangqinRen/PrivDNN

Problem

• Secure Multi-Party Computation
• Privacy-Preserving Deep Learning
• Model Privacy
• Data Privacy
• Encrypted DNN Evaluation

Innovation

• Secure Multi-Party Computation
• Partial DNN Encryption
• Privacy-Preserving Deep Learning
• Neuron Selection
• Efficient Encrypted Inference