Seoul National University's College of Engineering (Dean Byoungho Lee) announced on October 25 (Monday) that a research team led by Professor Jong-Seon No of the Department of Electrical and Computer Engineering has developed high-precision encryption technology to ensure data privacy in artificial intelligence.

Professor Jong-Seon No's research team announced a high-precision bootstrapping algorithm that enables an arbitrary of homomorphic operations in complete homomorphic encryption through Eurocrypt 2021, which was held on October 18 (Monday), Korean time.
 
Seoul National University's College of Engineering reported that this technology has prepared for an epochal turning point in which encrypted data can be applied to commercial artificial intelligence systems to provide a breakthrough in ensuring data privacy.
 
Eurocrypt, along with Crypto, is one of the world's top international academic conferences in the field of cryptography and until last year, a total of three papers were published by domestic researchers over the past five years.
 
During this Eurocrypt 2021, Professor Jong-Seon No's research team was the only one in Korea to have published a thesis, and in this paper, along with Professor Jong-sun Roh, researchers Joonwoo Lee, Eun-Sang Lee, and Lee Yong-woo (currently affiliated with Samsung Advanced Institute of Technology) of Seoul National University and Professor ‪Young-Sik Kim of Chosun University participated in the joint research together.
 
The CKKS algorithm, which is a homomorphic cipher that can perform operations in an encrypted state, has attracted attention as an algorithm used in machine learning for artificial intelligence by supporting real number operation. However, until now, the accuracy of the bootstrapping technique, which is the core technology that makes the homomorphic operation of CKKS sustainable, was not sufficient, making it was difficult to perform many calculations that are enough to process deep neural network calculations.
 
Based on the results of this study, it was possible to increase the computational accuracy of ciphertexts by up to 1,000 times compared to the previous one, and it became possible to process CKKS homomorphic operations by as many times as times.
 
In addition, through the technology introduced this time, it has become possible to perform operations of ResNets and VGGNets, which are standard deep neural network artificial intelligence models, with encrypted data encrypted. In fact, Professor Jong-Seon No's research team applied the results of this study to demonstrate the success of classification by artificial intelligence for homomorphic images in ResNet-20 for the first time in the world in collaboration with the Samsung Advanced Institute of Technology, and is also conducting research on encrypted artificial intelligence model learning.
 
Using this technology, it has become possible to develop or use artificial intelligence services while blocking personal information exposure by transmitting data containing highly sensitive personal information to an artificial intelligence server in an encrypted state.
 
A large amount of data is required to create an excellent artificial intelligence learning model, but it also contains a lot of sensitive data about individuals, which was considered a major obstacle to the development of good AI services.
 
This technology is expected to provide a breakthrough in the practical use of encryption technology that can guarantee data privacy by blocking information exposure in the development of artificial intelligence development.