Secure Inference

Secure inference aims to perform computations on sensitive data (like user inputs or model parameters) without revealing the data itself, addressing privacy concerns in machine learning applications. Current research focuses on optimizing secure inference for large language models (LLMs) and convolutional neural networks (CNNs) using techniques like secure multi-party computation (MPC), homomorphic encryption (HE), and quantization, often targeting specific bottlenecks like nonlinear activation functions or large linear layers. These advancements are crucial for enabling the widespread adoption of privacy-preserving machine learning in various sectors, including healthcare, finance, and collaborative research projects, by mitigating the risks associated with sensitive data exposure during model deployment.

Papers

September 9, 2024

CipherDM: Secure Three-Party Inference for Diffusion Model Sampling
Xin Zhao, Xiaojun Chen, Xudong Chen, He Li, Tingyu Fan, Zhendong Zhao
Diffusion Posterior Sampling Secure Multi Party Computation Secure Inference Private Data Analysis Multi Party Computation

August 7, 2024

MPC-Minimized Secure LLM Inference
Deevashwer Rathee, Dacheng Li, Ion Stoica, Hao Zhang, Raluca Popa
Scientific Inference LLM Inference Secure Multi Party Computation Inference Service Secure Inference

May 9, 2024

Ditto: Quantization-aware Secure Inference of Transformers upon MPC
Haoqi Wu, Wenjing Fang, Yancheng Zheng, Junming Ma, Jin Tan, Yinggui Wang, Lei Wang
Transformer Megatron Decepticons Transformer Inference Secure Inference Multi Party Computation Quantization Aware Knowledge Distillation

February 13, 2024

Privacy-Preserving Language Model Inference with Instance Obfuscation
Yixiang Yao, Fei Wang, Srivatsan Ravi, Muhao Chen
Obfuscation Technique Privacy Preserving Representation Language Model Inference Secure Inference Privacy Decision

February 1, 2024

Institutional Platform for Secure Self-Service Large Language Model Exploration
V. K. Cody Bumgardner, Mitchell A. Klusty, W. Vaiden Logan, Samuel E. Armstrong, Caylin Hickey, Jeff Talbert
Large Language Model Language Model Artificial Intelligence Learning Platform Secure Inference LoRA Module

December 28, 2023

SentinelLMs: Encrypted Input Adaptation and Fine-tuning of Language Models for Private and Secure Inference
Abhijit Mishra, Mingda Li, Soham Deo
Language Model Fine Tuning Pre Trained Language Model Neural Language Model Fine Tuned Language Model Token Embeddings Secure Inference Input Adaptation

November 15, 2023

Converting Transformers to Polynomial Form for Secure Inference Over Homomorphic Encryption
Itamar Zimerman, Moran Baruch, Nir Drucker, Gilad Ezov, Omri Soceanu, Lior Wolf
Transformer Megatron Decepticons Transformer Architecture Transformer Model Homomorphic Encryption Polynomial System Secure Inference

October 12, 2023

AutoFHE: Automated Adaption of CNNs for Efficient Evaluation over FHE
Wei Ao, Vishnu Naresh Boddeti
CNN Model Efficient Evaluation Polynomial Activation Polynomial Approximation Secure Inference

July 3, 2023

Pareto-Secure Machine Learning (PSML): Fingerprinting and Securing Inference Serving Systems
Debopam Sanyal, Jui-Tse Hung, Manav Agrawal, Prahlad Jasti, Shahab Nikkhoo, Somesh Jha, Tianhao Wang, Sibin Mohan, Alexey Tumanov
Model Extraction Attack Model Deployment Secure Inference Fingerprinting Lemma Secure Machine Learning Victim Model

June 11, 2023

Efficient Skip Connections Realization for Secure Inference on Encrypted Data
Nir Drucker, Itamar Zimerman
Homomorphic Encryption Deep Learning Application Skip Connection Secure Inference Cryptographic Building Block

September 14, 2022

SEEK: model extraction attack against hybrid secure inference protocols
Si Chen, Junfeng Fan
Homomorphic Encryption Model Extraction Attack Inference Service Party Computation Secure Inference

August 26, 2022

Efficient ML Models for Practical Secure Inference
Vinod Ganesan, Anwesh Bhattacharya, Pratyush Kumar, Divya Gupta, Rahul Sharma, Nishanth Chandran
Scientific Inference Efficient Machine Learning Secure Inference FoldFlow 2

May 13, 2022

Collaborative Drug Discovery: Inference-level Data Protection Perspective
Balazs Pejo, Mina Remeli, Adam Arany, Mathieu Galtier, Gergely Acs
Drug Discovery Privacy Risk Inference Attack Model Collaboration Drug Candidate Secure Inference

March 5, 2022

Tabula: Efficiently Computing Nonlinear Activation Functions for Secure Neural Network Inference
Maximilian Lam, Michael Mitzenmacher, Vijay Janapa Reddi, Gu-Yeon Wei, David Brooks
Activation Function Lookup Table Real Time Inference Secure Multi Party Computation Secure Inference Secure Deep

February 27, 2022

Split HE: Fast Secure Inference Combining Split Learning and Homomorphic Encryption
George-Liviu Pereteanu, Amir Alansary, Jonathan Passerat-Palmbach
Neural Network Homomorphic Encryption Split Learning Membership Inference Model Extraction Attack Learning Benchmark Secure Inference