News
Computing professor develops a framework with collaborators for enhancing data privacy for AI model training
December 06, 2023
Dr. Yuan Cheng, associate professor of computing, recently published a paper with his former master’s student Stephen Ly, and faculty collaborators from California State University, Sacramento at the 20th International Conference on Privacy, Security, and Trust (PST2023), Copenhagen, Denmark, August 21-23, 2023.
The paper, titled "A Secure Distributed Learning Framework Using Homomorphic Encryption," presents a method for secure and efficient AI model training using homomorphic encryption. By combining distributed training with the CKKS homomorphic encryption scheme, it speeds up training time while preserving data privacy. Experimental results show reduced runtime and competitive accuracy on the MNIST benchmarking dataset, making it a practical solution for secure AI training.
This continues Dr. Cheng's work in the security and privacy of emerging technologies and marks his second venture into artificial intelligence (AI). In his previous work, he leveraged AI techniques to enhance the accuracy of targeted password and passphrase predictions.