USENIX Security2025Top-tier venue
SLOTHE : Lazy Approximation of Non-Arithmetic Neural Network Functions over Encrypted Data
Abstract
Machine Learning as a Service (MLaaS) introduces strong privacy concerns for both clients and model providers. Fully Homomorphic Encryption (FHE) offers a promising solution by enabling inference over encrypted data, but its limited expressiveness requires approximating non-arithmetic functions (NAFs) with polynomials, often leading to significant accuracy and performance trade-offs. Existing works adopt an eager approximation (EA) strategy, which statically replaces each NAF with a fixed polynomial, locking in computational errors and limiting optimization opportunities. We propose SLOTHE, a lazy approximation (LA) solution that recursively decomposes NAF codes into arithmetic and nonarithmetic sub-functions, selectively approximating only the non-arithmetic components when required. SLOTHE introduces a tunable cost model to balance accuracy and latency, and incorporates FHE-aware optimizations to eliminate redundant computation. Implemented using CKKS, SLOTHE achieves up to 42,378× lower maximum error than EA-based works, with improved inference accuracy and latency across BERT-based transformers. SLOTHE can also be adapted for MPC-based protocols, making it a flexible tool for secure neural network inference.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
1 top-tier paper has cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Iron: Private Inference on Transformers cites CrypTen: Secure Multi-Party Computation Meets Machine Learning
- Iron: Private Inference on Transformers cites GAZELLE: A Low Latency Framework for Secure Neural Network Inference
- Low-Complexity Deep Convolutional Neural Networks on Fully Homomorphic Encryption Using Multiplexed Parallel Convolutions cites GAZELLE: A Low Latency Framework for Secure Neural Network Inference
- CrypTen: Secure Multi-Party Computation Meets Machine Learning cites GAZELLE: A Low Latency Framework for Secure Neural Network Inference
- RBOOT: Accelerating Homomorphic Neural Network Inference by Fusing ReLU within Bootstrapping cites Low-Complexity Deep Convolutional Neural Networks on Fully Homomorphic Encryption Using Multiplexed Parallel Convolutions
- RBOOT: Accelerating Homomorphic Neural Network Inference by Fusing ReLU within Bootstrapping cites GAZELLE: A Low Latency Framework for Secure Neural Network Inference
Cited by top-tier papers1
Builds on29
- ALBERT: A Lite BERT for Self-supervised Learning of Language RepresentationsZhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel et al.ICLR 2020 · 7,418 citations
- GAZELLE: A Low Latency Framework for Secure Neural Network InferenceChiraag Juvekar, Vinod Vaikuntanathan, Anantha P. ChandrakasanUSENIX Security 2018 · 1,075 citations
- CrypTen: Secure Multi-Party Computation Meets Machine LearningBrian Knott, Shobha Venkataraman, Awni Y. Hannun, Shubho Sengupta et al.NeurIPS 2021 · 573 citations
- Iron: Private Inference on TransformersMeng Hao, Hongwei Li, Hanxiao Chen, Pengzhi Xing et al.NeurIPS 2022 · 209 citations
- Low-Complexity Deep Convolutional Neural Networks on Fully Homomorphic Encryption Using Multiplexed Parallel ConvolutionsEunsang Lee, Joon-Woo Lee, Junghyun Lee, Young-Sik Kim et al.ICML 2022 · 171 citations
Related papers
- Falcon: Fast Spectral Inference on Encrypted DataQian Lou, Wen-jie Lu, Cheng Hong, Lei JiangNeurIPS 2020 · 50 citations
- HEPIC: Private Inference over Homomorphic Encryption with Client InterventionKevin Nam, Youyeon Joo, Seungjin Ha, Hyungon Moon et al.ASPLOS 2026
- Fast and Private Inference of Deep Neural Networks by Co-designing Activation FunctionsAbdulrahman Diaa, Lucas Fenaux, Thomas Humphries, Marian Dietz et al.USENIX Security 2024 · 13 citations
- EncryptedLLM: Privacy-Preserving Large Language Model Inference via GPU-Accelerated Fully Homomorphic EncryptionLeo de Castro, Daniel Escudero, Adya Agrawal, Antigoni Polychroniadou et al.ICML 2025
- From Individual Computation to Allied Optimization: Remodeling Privacy-Preserving Neural Inference with Function Input TuningQiao Zhang, Tao Xiang, Chunsheng Xin, Hongyi WuS&P 2024 · 12 citations
