# PrivCirNet: Efficient Private Inference via Block Circulant Transformation

arXiv: https://arxiv.org/abs/2405.14569 · HTML: https://mlsanity.com/paper/2405.14569/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 58 references that are in the map.
- First posted on arXiv 23 May 2024.

## Builds on (most cited first, 18 of 18)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstruction](https://mlsanity.com/paper/2102.05426/) (2021)
- [Escaping the Big Data Paradigm with Compact Transformers](https://mlsanity.com/paper/2104.05704/) (2021)
- [HAWQ-V2: Hessian Aware trace-Weighted Quantization of Neural Networks](https://mlsanity.com/paper/1911.03852/) (2019)
- [MPCFormer: fast, performant and private Transformer inference with MPC](https://mlsanity.com/paper/2211.01452/) (2022)
- [DeepReDuce: ReLU Reduction for Fast Private Inference](https://mlsanity.com/paper/2103.01396/) (2021)
- [SHE: A Fast and Accurate Deep Neural Network for Encrypted Data](https://mlsanity.com/paper/1906.00148/) (2019)
- [Selective Network Linearization for Efficient Private Inference](https://mlsanity.com/paper/2202.02340/) (2022)
- [MPCViT: Searching for Accurate and Efficient MPC-Friendly Vision Transformer with Heterogeneous Attention](https://mlsanity.com/paper/2211.13955/) (2022)
- [Learning to Linearize Deep Neural Networks for Secure and Efficient Private Inference](https://mlsanity.com/paper/2301.09254/) (2023)
- [GALA: Greedy ComputAtion for Linear Algebra in Privacy-Preserved Neural Networks](https://mlsanity.com/paper/2105.01827/) (2021)
- [HyPHEN: A Hybrid Packing Method and Optimizations for Homomorphic Encryption-Based Neural Networks](https://mlsanity.com/paper/2302.02407/) (2023)
- [CoPriv: Network/Protocol Co-Optimization for Communication-Efficient Private Inference](https://mlsanity.com/paper/2311.01737/) (2023)
- [NeuJeans: Private Neural Network Inference with Joint Optimization of Convolution and FHE Bootstrapping](https://mlsanity.com/paper/2312.04356/) (2023)
- [Falcon: Accelerating Homomorphically Encrypted Convolutions for Efficient Private Mobile Network Inference](https://mlsanity.com/paper/2308.13189/) (2023)
- [Toward Practical Privacy-Preserving Convolutional Neural Networks Exploiting Fully Homomorphic Encryption](https://mlsanity.com/paper/2310.16530/) (2023)
- [Impala: Low-Latency, Communication-Efficient Private Deep Learning Inference](https://mlsanity.com/paper/2205.06437/) (2022)
- [Efficient Pruning for Machine Learning Under Homomorphic Encryption](https://mlsanity.com/paper/2207.03384/) (2022)

## Built on by (most cited first, 7 of 7)

- [CipherPrune: Efficient and Scalable Private Transformer Inference](https://mlsanity.com/paper/2502.16782/) (2025)
- [An Efficient Private GPT Never Autoregressively Decodes](https://mlsanity.com/paper/2505.15252/) (2025)
- [Privacy-Preserving Semantic Segmentation from Ultra-Low-Resolution RGB Inputs](https://mlsanity.com/paper/2507.16034/) (2025)
- [PRISM: Sensitivity-Aware PolynoMial PRuning for EffIcient Neural Network Encryption](https://mlsanity.com/paper/2607.18342/) (2026)
- [Gecko: Fast Private Inference via Secure Public Encoder Offloading](https://mlsanity.com/paper/2608.02378/) (2026)
- [OptiPrime: Optimizing Private Inference through Protocol-Hardware Co-design](https://mlsanity.com/paper/2609.16898/) (2026)
- [Fiona: Accelerating FHE Inference with Packing-Aware Ternary Weights](https://mlsanity.com/paper/2609.35352/) (2026)

## Abstract (arXiv)

Homomorphic encryption (HE)-based deep neural network (DNN) inference protects data and model privacy but suffers from significant computation overhead. We observe transforming the DNN weights into circulant matrices converts general matrix-vector multiplications into HE-friendly 1-dimensional convolutions, drastically reducing the HE computation cost. Hence, in this paper, we propose method, a protocol/network co-optimization framework based on block circulant transformation. At the protocol level, PrivCirNet customizes the HE encoding algorithm that is fully compatible with the block circulant transformation and reduces the computation latency in proportion to the block size. At the network level, we propose a latency-aware formulation to search for the layer-wise block size assignment based on second-order information. PrivCirNet also leverages layer fusion to further reduce the inference cost. We compare PrivCirNet with the state-of-the-art HE-based framework Bolt (IEEE S&P 2024) and the HE-friendly pruning method SpENCNN (ICML 2023). For ResNet-18 and Vision Transformer (ViT) on Tiny ImageNet, PrivCirNet reduces latency by 5.0× and 1.3× with iso-accuracy over Bolt, respectively, and improves accuracy by 4.1% and 12% over SpENCNN, respectively. For MobileNetV2 on ImageNet, PrivCirNet achieves 1.7× lower latency and 4.2% better accuracy over Bolt and SpENCNN, respectively. Our code and checkpoints are available on Git Hub.