Please note: This PhD seminar will take place in DC 3317 and online.
Yiwei Lu, PhD candidate
David R. Cheriton School of Computer Science
Supervisors: Professors Yaoliang Yu, Sun Sun
Advanced generative models, particularly latent diffusion models, have demonstrated remarkable capabilities in producing high-quality images. However, recent research reveals a concerning trend: these models often simply memorize training data, potentially reproducing copyrighted content from their training sets and thereby infringing on the rights of human artists.
This talk introduces an even more significant threat: a concealed form of copyright infringement where an adversary can reproduce copyrighted material without direct access to it. The presentation is divided into two parts. First, we explore this hidden copyright infringement threat and its implications. Then, we examine machine unlearning as a potential solution for removing copyrighted data from training sets. Our discussion focuses on methods for data owners to audit and verify the effectiveness of the unlearning process.
To attend this PhD seminar in person, please go to DC 3317. You can also attend virtually on Zoom.