On Detecting Fake (AI-generated) Images

(Please note this talk starts at 12:00pm (not 12:15pm) in Engineering Hall Room 2349 (not CS 1240).  This is a one-time change.)

You are cordially invited to the weekly CS Machine Learning Lunch Meetings. This is a chance to get to know machine learning professors, and talk to your fellow researchers.  This week's meeting will be on Tuesday Feb. 14 12:00-1:00pm in Engineering Hall Room 2349. Professor Yong Jae Lee will tell us about detecting AI generated images, see abstract below.

Abstract: Generative models have made tremendous advances in the past few years.  And the field is moving incredibly fast -- it seems to take only a couple of weeks until we hear about the latest and greatest image generation model.  While exciting, there are also negative societal aspects that can arise from the fake visual content generated by AI models.  In this talk, I will present our lab's ongoing work on detecting fake, AI-generated images.  What makes the problem particularly challenging, and interesting, is that there are many different types of AI models that can produce fake images (e.g., GANs, VAEs, diffusion models).  I'll discuss our initial findings towards our goal of creating a general-purpose fake image detector, and present some open questions.  I would like this talk to be interactive, especially since it is based on ongoing, incomplete work.

If you would like to be informed of future CS Machine Learning Lunch Meetings, please sign up our mailing list at https://lists.cs.wisc.edu/mailman/listinfo/mllm -- please use your cs or wisc email.  After you enter your email, the system will send you an email for confirmation.  Only after you respond to that email will you be on the mailing list.