Private Inference

Private inference (PI) aims to perform computations on encrypted data, protecting both user data and model parameters during machine learning inference. Current research focuses on improving the efficiency of PI, particularly for large language models (LLMs) and vision transformers (ViTs), by optimizing communication, reducing computational overhead (e.g., through approximation of non-linear functions like ReLU), and developing novel algorithms like adaptive PI and layer-wise approximation techniques. These advancements are crucial for enabling widespread adoption of privacy-preserving machine learning in cloud-based services and other sensitive applications.

Papers

October 16, 2024

AERO: Softmax-Only LLMs for Efficient Private Inference
Nandan Kumar Jha, Brandon Reagen
Softmax Function Private Inference Aircraft Design Decoder Only Language Model Proprietary Large Language Model

October 12, 2024

PrivQuant: Communication-Efficient Private Inference with Quantized Network/Protocol Co-Optimization
Tianshi Xu, Shuzhang Zhong, Wenxuan Zeng, Runsheng Wang, Meng Li
Neural Network Network Programming Inference Strategy Private Inference Co Optimization

July 8, 2024

AdaPI: Facilitating DNN Model Adaptivity for Efficient Private Inference in Edge Computing
Tong Zhou, Jiahui Zhao, Yukui Luo, Xi Xie, Wujie Wen, Caiwen Ding, Xiaolin Xu
Edge Computing Private Inference

May 25, 2024

FastQuery: Communication-efficient Embedding Table Query for Private LLM Inference
Chenqi Lin, Tianshi Xu, Zebin Yang, Runsheng Wang, Ru Huang, Meng Li
Homomorphic Encryption User Privacy Private Inference Query Optimization Table Question

May 24, 2024

Comet: A Communication-efficient and Performant Approximation for Private Transformer Inference
Xiangrui Xu, Qiao Zhang, Rui Ning, Chunsheng Xin, Hongyi Wu
Transformer Model Transformer Based Model Transformer Inference Private Inference Inference Performance

May 23, 2024

PrivCirNet: Efficient Private Inference via Block Circulant Transformation
Tianshi Xu, Lemeng Wu, Runsheng Wang, Meng Li
Deep Neural Network Homomorphic Encryption Private Inference Block Size Circulant Matrix

February 8, 2024

Linearizing Models for Efficient yet Robust Private Inference
Sreetama Sarkar, Souvik Kundu, Peter A. Beerel
High Efficiency Reverse Graph Linearization Private Inference Weight Mask Private Bayesian Model Latency ReLU Operation

December 30, 2023

ConfusionPrompt: Practical Private Inference for Online Large Language Models
Peihua Mai, Youjia Yang, Ran Yan, Rui Ye, Yan Pang
Large Language Model Privacy Amplification Black Box Large Language Model Private Inference Group Privacy

December 23, 2023

Regularized PolyKervNets: Optimizing Expressiveness and Efficiency for Private Inference in Deep Neural Networks
Toluwani Aremu
Deep Neural Network High Efficiency Expressive Speech Private Inference Polynomial Activation Privacy Preserving Computation

October 16, 2023

Optimized Layerwise Approximation for Efficient Private Inference on Fully Homomorphic Encryption
Junghyun Lee, Eunsang Lee, Young-Sik Kim, Yongwoo Lee, Joon-Woo Lee, Yongjune Kim, Jong-Seon No
Homomorphic Encryption Private Inference Optimization Layer Efficient Approximation Polynomial Approximation

October 13, 2023

Split-and-Denoise: Protect large language model inference with local differential privacy
Peihua Mai, Ran Yan, Zhe Huang, Youjia Yang, Yan Pang
Text Embeddings Local Differential Privacy Large Language Model Inference Private Inference Output Embeddings

October 6, 2023

PriViT: Vision Transformers for Fast Private Inference
Naren Dhyani, Jianqiao Mo, Minsu Cho, Ameya Joshi, Siddharth Garg, Brandon Reagen, Chinmay Hegde
Vision Transformer State of the Art Deep Computer Vision Application Private Inference

September 9, 2023

Approximating ReLU on a Reduced Ring for Efficient MPC-based Private Inference
Kiwan Maeng, G. Edward Suh
High Efficiency Privacy Sensitive Private Inference Non Private Multi Party Computation

August 20, 2023

AutoReP: Automatic ReLU Replacement for Fast Private Network Inference
Hongwu Peng, Shaoyi Huang, Tong Zhou, Yukui Luo, Chenghong Wang, Zigeng Wang, Jiahui Zhao, Xi Xie, Ang Li, Tony Geng, Kaleel Mahmood, Wujie Wen, Xiaolin Xu, Caiwen Ding
ImageNet Dataset ReLU Layer Tiny ImageNet Private Inference Polynomial Approximation ReLU Operation

July 9, 2023

Towards Fast and Scalable Private Inference
Jianqiao Mo, Karthik Garimella, Negar Neda, Austin Ebel, Brandon Reagen
Homomorphic Encryption Encryption Method Private Inference Privacy Preserving Computation Oblivious Transfer

May 28, 2023

LLMs Can Understand Encrypted Prompt: Towards Privacy-Computing Friendly Transformers
Xuanqi Liu, Zhuotao Liu
Large Language Model Private Data Inference Cost Transformer Based Large Language Model Prompt Attack Private Inference Privacy Preserving Inference Thought Empowers Transformer

May 25, 2023

MERGE: Fast Private Text Generation
Zi Liang, Pinghui Wang, Ruofei Zhang, Nuo Xu, Lifeng Xing, Shuo Zhang
Language Model Language Generation Transformer Based Language Model Private Inference Private INFORMATION RETRIEVAL

April 26, 2023

Making Models Shallow Again: Jointly Learning to Reduce Non-Linearity and Depth for Latency-Efficient Private Inference
Souvik Kundu, Yuke Zhang, Dake Chen, Peter A. Beerel
Large Depth Convolution Layer Convolutional Block ReLU Layer Private Inference Shallow Layer Non Linearity ReLU Function

January 23, 2023

Learning to Linearize Deep Neural Networks for Secure and Efficient Private Inference
Souvik Kundu, Shunlin Lu, Yuke Zhang, Jacqueline Liu, Peter A. Beerel
Neural Network LeArning Abstract Deep Neural Network ReLU Layer Secure Approach Private Inference ReLU Function

November 2, 2022

MPCFormer: fast, performant and private Transformer inference with MPC
Dacheng Li, Rulin Shao, Hongyi Wang, Han Guo, Eric P. Xing, Hao Zhang
Inference Latency Transformer Inference Private Inference Inference Service Multi Party Computation Sl MPC

Private Inference

Papers

AERO: Softmax-Only LLMs for Efficient Private Inference

PrivQuant: Communication-Efficient Private Inference with Quantized Network/Protocol Co-Optimization

AdaPI: Facilitating DNN Model Adaptivity for Efficient Private Inference in Edge Computing

FastQuery: Communication-efficient Embedding Table Query for Private LLM Inference

Comet: A Communication-efficient and Performant Approximation for Private Transformer Inference

PrivCirNet: Efficient Private Inference via Block Circulant Transformation

Linearizing Models for Efficient yet Robust Private Inference

ConfusionPrompt: Practical Private Inference for Online Large Language Models

Regularized PolyKervNets: Optimizing Expressiveness and Efficiency for Private Inference in Deep Neural Networks

Optimized Layerwise Approximation for Efficient Private Inference on Fully Homomorphic Encryption

Split-and-Denoise: Protect large language model inference with local differential privacy

PriViT: Vision Transformers for Fast Private Inference

Approximating ReLU on a Reduced Ring for Efficient MPC-based Private Inference

AutoReP: Automatic ReLU Replacement for Fast Private Network Inference

Towards Fast and Scalable Private Inference

LLMs Can Understand Encrypted Prompt: Towards Privacy-Computing Friendly Transformers

MERGE: Fast Private Text Generation

Making Models Shallow Again: Jointly Learning to Reduce Non-Linearity and Depth for Latency-Efficient Private Inference

Learning to Linearize Deep Neural Networks for Secure and Efficient Private Inference

MPCFormer: fast, performant and private Transformer inference with MPC