First Demonstration of Unclonable Double Encryption 28nm RRAM-Based Compute-in-Memory Macro for Confidential AI

Yiyang Chen,Lixia Han, Ao Shi, Lianliang Wu, Hairuo Lu, Haozhang Yang, Jiaqi Li, Yulin Feng,Zheng Zhou,Lifeng Liu,Xiaoyan Liu,Jinfeng Kang,Peng Huang

2024 IEEE International Electron Devices Meeting (IEDM)（2024）

Cited 0|Views13

Abstract

For the first time, we demonstrate an unclonable 28nm resistive random-access memory (RRAM) based Compute-in-Memory (CIM) macro equipped with in-situ double encryption and matrix-vector multiplication (MVM) to protect the AI models and avoid the risk of exposing sensitive user data. Main features include: 1) The sign and the value of the neural network weights are encrypted with different keys, which shows 2960 times improvement in key space compared with previous encryption CIM. 2) The multi-bit weights can be in-situ decrypted with the keys within one computing cycle, enabling the model based on our macro to achieve 50% improvement in measured energy efficiency compared with encryption CIM. 3) The variability of the MOSFET creates an unclonable identity for each macro. Therefore, the created macro is still trustworthy even when all the keys are exposed, which is beyond the capability of previous encryption methods.

Translated text

Key words

Double Encryption,Energy Efficiency,Matrix Multiplication,Resistive Random Access Memory,Key Space,Encryption Method,AI Models,Voltage-gated,Plaintext,Area Overhead,Key Exposure,Sign Bit,Physical Unclonable Functions

求助PDF

上传PDF

Bibtex

AI Read Science

AI Summary

AI Summary is the key point extracted automatically understanding the full text of the paper, including the background, methods, results, conclusions, icons and other key content, so that you can get the outline of the paper at a glance.

Example

Background

Key content

Introduction

Methods

Results

Related work

Fund

Key content

Pretraining has recently greatly promoted the development of natural language processing (NLP)
We show that M6 outperforms the baselines in multimodal downstream tasks, and the large M6 with 10 parameters can reach a better performance
We propose a method called M6 that is able to process information of multiple modalities and perform both single-modal and cross-modal understanding and generation
The model is scaled to large model with 10 billion parameters with sophisticated deployment, and the 10 -parameter M6-large is the largest pretrained model in Chinese
Experimental results show that our proposed M6 outperforms the baseline in a number of downstream tasks concerning both single modality and multiple modalities We will continue the pretraining of extremely large models by increasing data to explore the limit of its performance

Upload PDF to Generate Summary

Must-Reading Tree

Example

Generate MRT to find the research sequence of this paper

Data Disclaimer

The page data are from open Internet sources, cooperative publishers and automatic analysis results through AI technology. We do not make any commitments and guarantees for the validity, accuracy, correctness, reliability, completeness and timeliness of the page data. If you have any questions, please contact us by email: report@aminer.cn

Chat Paper

Summary is being generated by the instructions you defined