Changho Shin

(신창호, @ch-shin)

Face

Ph.D. Student in CS at University of Wisconsin-Madison

email cshin23@wisc.edu
Twitter
Github
insert_drive_file CV

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.

News

Conference Publications

[C7] Weak-to-Strong Generalization Through the Data-Centric Lens, ICLR 2025
Changho Shin, John Cooper, Frederic Sala

[C6] Personalize Your LLM: Fake it then Align it, NAACL 2025 Findings
Yijing Zhang, Dyah Adila, Changho Shin, Frederic Sala

[C5] OTTER: Improving Zero-Shot Classification via Optimal Transport, NeurIPS 2024
Changho Shin, Jitian Zhao, Sonia Cromp, Harit Vishwakarma, Frederic Sala

[C4] Zero-Shot Robustification of Zero-Shot Models, ICLR 2024
Also presented in NeurIPS 2023 R0-FoMo Workshop (Best Paper Award Honorable Mention)
Dyah Adila*, Changho Shin*, Linrong Cai, Frederic Sala

[C3] Mitigating Source Bias for Fairer Weak Supervision, NeurIPS 2023
Changho Shin, Sonia Cromp, Dyah Adila, Frederic Sala

[C2] Universalizing Weak Supervision, ICLR 2022
Changho Shin, Winfred Li, Harit Vishwakarma, Nicholas Roberts, Frederic Sala

[C1] Subtask Gated Networks for Non-Intrusive Load Monitoring, AAAI 2019
Changho Shin, Sunghwan Joo, Jaeryun Yim, Hyoseop Lee, Taesup Moon, and Wonjong Rhee

Journal Publications

[J2] The ENERTALK dataset, 15 Hz electricity consumption data from 22 houses in Korea, Scientific Data
Changho Shin, Eunjung Lee, Jeongyun Han, Jaeryun Yim, Hyoseop Lee, and Wonjong Rhee

[J1] Data Requirements for Applying Machine Learning to Energy Disaggregation, Energies
Changho Shin, Seungeun Rho, Hyoseop Lee, and Wonjong Rhee

Workshop Publications

[W5] Is Free Self-Alignment Possible?, NeurIPS 2024 MINT Workshop
Dyah Adila, Changho Shin, Yijing Zhang, Frederic Sala

[W4] Foundation Models Can Robustify Themselves, For Free, NeurIPS 2023 R0-FoMo Workshop (Best Paper Award Honorable Mention)
Dyah Adila*, Changho Shin*, Linrong Cai, Frederic Sala

[W3] Pool-Search-Demonstrate: Improving Data-wrangling LLMs via better in-context examples , NeurIPS 2023 TRL Workshop (Oral)
Changho Shin*, Joon Suk Huh*, Elina Choi

[W2] Multimodal Data Curation via Object Detection and Filter Ensembles, ICCV 2023 DataComp Workshop (Filtering Track Rank #1 (Small))
Changho Shin*, Tzu-heng Huang*, Sui Jiet Tay, Dyah Adila, Frederic Sala

[W1] Can we get smarter than majority vote? Efficient use of individual rater’s labels for content moderation, NeurIPS 2022 ENLSP Workshop
Changho Shin, Alice Schoenauer Sebag

AWARDS

Work Experience

apartment Microsoft Research, Cambridge, MA 2025.06 - 2025.08

apartment Snorkel AI, Redwood City, CA (Remote) 2024.06 - 2024.08

apartment Twitter, San Francisco, CA 2022.06 - 2022.08

apartment Encored Technologies, Seoul, Korea 2018.01 - 2020.07

apartment KIDA (Korea Institute for Defense Analyses), Seoul, Korea 2017.01 - 2017.12

Teaching Experience

school University of Wisconsin-Madison 2020.09 -