Research talk

Universalizing weak supervision

June 07, 2024
| 12:00 PM - 1:00 PM Pacific Time

Watch on demand

By submitting this form, I agree to the Terms of Use and acknowledge that my information will be used in accordance with the Privacy Policy.

During this research talk, you’ll see how researchers are pushing the boundaries of what’s possible with weak supervision.

PhD Student Changho Shin from the University of Wisconsin-Madison will discuss how a new weak supervision framework allows users to apply the technique across task types without extensive customization—and even extend weak supervision to previously inaccessible label types, such as ranking.

The talk will address:

  • How weak supervision scales labeling efforts.
  • How this new framework eases transitions between task types.
  • How the new framework allows weak supervision to apply to new task types.

Presented by

Image

Changho Shin

Postdoctoral Scholar at Princeton University
Princeton University

I am a final-year PhD student in Computer Science at University of Wisconsin-Madison, where I am fortunate to be advised by Frederic Sala. Before that, I was a master’s student at Seoul National University, where I was lucky to learn deep learning, exploratory data analysis, and information theory from Wonjong Rhee. Prior to that, I received B.A in Psychology and B.S. in Computer Science and Engineering from Seoul National University.

My research focuses on data-centric AI, particularly programmatic weak supervision and weak-to-strong generalization in foundation models. These approaches use weaker models as supervision sources to train stronger models, providing labels, reward signals, and verification signals that guide more capable systems. I have also explored inference-time steering, which involves intervening on internal representations to improve robustness, alignment, and personalization of foundation models at inference time without fine-tuning.

Looking ahead, my vision is to develop strategies for supervising superhuman-level intelligence, where traditional human oversight is no longer sufficient. My research currently focuses on two directions. The first is weak-to-strong generalization, where weaker models are used as supervision sources to train stronger ones, providing labels, reward signals, and verification for more capable systems. The second is out-of-distribution (OOD) generalization, including challenges such as easy-to-hard generalization, length generalization, and compositional generalization.