| Project Reference: | ITP/006/22LP |
| Project Title: | Efficient Logistic Regression on Encrypted Data for Privacy Preserving Data Sharing |
| Hosting Institution: | LSCM R&D Centre (LSCM) |
| Abstract: | With the increasing adoption of big data technologies, artificial intelligence, and the Internet of Things, data sharing becomes crucial for organizations across the public and private sectors globally. Gartner, in the article "Data is a Business Necessity to Accelerate Digital Business", predicts that by 2023, organizations that promote data sharing will outperform those who do not on most business value metrics. Data sharing can also help to deal with some of the society’s biggest challenges more innovatively and effectively. One of the major difficulties data trading platforms face is to protect data owners' interests by maintaining appropriate level of privacy and confidentiality through encryption technology and still can perform regulatory compliance due diligence check, and data mining activities to generate valuable insights. Logistic regression is one of the most common machine learning methods being used in the plaintext scenario to achieve these objectives. Baldwin and Krishna Dayanidhi in the book "Natural Language Processing with Java and LingPipe Cookbook", points out that logistic regression is one of the machine learning methods that is responsible for the majority of industrial classifier and is most certainly one of the best performing classifiers available. Furthermore, a neural network can be viewed as a series of logistic regression classifiers stacked on top of each other (see "Speech and Language Processing, 3rd edition, by Dainiel Jurasky and James Martin). We therefore propose to develop and implement an efficient method to perform logistic regression on encrypted data with appropriate accuracy and precision. This provides a foundation for the development of a privacy-preserving data sharing platform that can perform appropriate due diligence check and generate valuable insights. Furthermore, since all data remain encrypted throughout the process, confidentiality and privacy are achieved through encryption. The platform can also act as a privacy preserving data mining service provider to data owners. |
| Project Coordinator: | Dr Russell Siu Wai Yiu |
| Approved Funding Amount: | HK$ 2.76 M |
| Project Period: | 31 Mar 2022 - 30 Mar 2023 |
