Efficient encrypted deep learning inference with logic gate networks
EI-DDLGN: Efficient Encrypted Inference with Deep Differentiable Logic Gate Networks under TFHE
Cryptography and SecurityMachine Learning
Summary
Deep learning models can protect your data by doing their calculations while the data stays encrypted, but this usually takes a lot of time. The authors studied a new way to run these secure calculations by using logic gates, which computers use to handle simple true-or-false operations. Their method fits well with the encryption type they use and speeds up the process while keeping accuracy high. They tested this idea on common image and data sets and found it much faster than traditional methods without losing quality.
What this means in practice
- •For cloud service teams: Run encrypted deep learning predictions on sensitive data faster using logic gate networks without compromising privacy.
- •For security software developers: Implement efficient privacy-preserving AI models that reduce secure inference delays by avoiding costly arithmetic operations under TFHE.
Authors
Mahmoud Y. M. Yassin, Mahmoud AbdelHafeez Sayed, Mostafa Taha
Abstract
Privacy-preserving inference via Torus Fully Homomorphic Encryption (TFHE) provides strong protection for sensitive data in outsourced deep learning applications. However, most TFHE-compatible neural network frameworks remain based on arithmetic neural architectures, resulting in high inference latency due to programmable bootstrapping (PBS), accumulator growth, and circuit bit-width sensitivity. In this work, we investigate Deep Differentiable Logic Gate Networks (DDLGNs) as a Boolean-native alternative for encrypted inference under TFHE. Because DDLGNs learn Boolean computations directly and discretize into fixed logic gate networks, their inference procedure is naturally aligned with TFHE's Boolean execution model and avoids arithmetic accumulation in hidden layers. We present EI-DDLGN, the first in-depth study of TFHE-based DDLGN inference, and characterize how encrypted execution cost depends on model size, learned Boolean-function distribution, and propagated wire status. We also introduce Model-Fixed-Wire PBS Bypass (MFW-PBS Bypass), a semantics-preserving execution strategy that eliminates unnecessary PBS operations without modifying the learned network topology. Evaluations across 72 depth-width configurations on MNIST, FashionMNIST, and UCI Phishing show that DDLGNs constitute an efficient alternative to arithmetic TFHE inference, achieving substantially improved accuracy-latency trade-offs. Notably, on MNIST, EI-DDLGN-Small matches the accuracy of QAT-FCNN-4 while reducing encrypted inference latency by 13.4x. Our implementation is available at https://github.com/Carleton-SCI/EI-DDLGN