| Author: | Ding, Renjie |
| Title: | Privacy-preserving generative adversarial networks |
| Degree: | M.Sc. |
| Year: | 2023 |
| Department: | Department of Electrical and Electronic Engineering |
| Pages: | 1 volume (unpaged) : color illustrations |
| Language: | English |
| Abstract: | As the pervasive influence of artificial intelligence extends across diverse domains, the demand for generative models has surged, finding applications in text generation, style transfer, and super-resolution. However, the training of these models, particularly Generative Adversarial Networks (GANs), requires extensive datasets that may inadvertently contain sensitive personal information. Safeguarding this privacy information from potential differential attacks becomes imperative. This research project is dedicated to exploring the integration of differential privacy into adversarial networks, focusing on label differential privacy and leveraging the random response technique. The primary objective revolves around image generation tasks. Implemented within this study are two prominent adversarial models: Wasserstein GAN (WGAN) and Auxiliary Classifier GAN (ACGAN). Given the distinct characteristics of these adversarial models, the approach to label processing in random response exhibits nuanced variations. Through a meticulous examination of these techniques, this project contributes to the broader understanding of preserving privacy in the context of generative models. |
| Rights: | All rights reserved |
| Access: | restricted access |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 8271.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 583.77 kB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/13868