Scholar
Yeda Song
Google Scholar ID: 6bjVlTsAAAAJ
Ph.D. student in Computer Science and Engineering, University of Michigan
Citations & Impact
Citations
37
H-index
2
i10-index
1
Publications
5
Co-authors
9
list available
Publications
Resume (English only)
Academic Achievements
- “Scalable Video-to-Dataset Generation for Cross-Platform Mobile Agents,” CVPR 2025 & CVPR Workshop on Multimodal Foundation Models 2025; introduced MONDAY dataset (313K annotated frames from 20K instructional videos) enabling robust cross-platform mobile agent generalization
- “Mobile OS Task Procedure Extraction from YouTube,” NeurIPS Workshop on Video-Language Models 2024 (Non-Archival); proposed MOTIFY method for extracting task sequences from YouTube without manual annotation
- “Compositional Conservatism: A Transductive Approach in Offline Reinforcement Learning,” ICLR 2024; proposed COCOA to address distributional shifts in offline RL
- “MPChat: Towards Multimodal Persona-Grounded Conversation,” ACL 2023; constructed MPChat multimodal persona-grounded dialogue dataset, demonstrating critical role of visual modality
- Rackham Conference Travel Grant (Dec. 2024)
- AI Fellowship (Mar. 2022 – Feb. 2024)
- Presidential Science Scholarship (Mar. 2017 – Feb. 2021)
- Merit-Based Undergraduate Scholarship (Mar. 2017 – Jun. 2017)
- Hanseong Nobel Scholarship (Mathematics Sector, Mar. 2015 – Feb. 2017)
Education
- Ph.D. student in Computer Science & Engineering, University of Michigan (Aug. 2024 – Present), advised by Prof. Honglak Lee
- M.S. in Artificial Intelligence, Seoul National University (Mar. 2022 – Feb. 2024), advised by Prof. Gunhee Kim in the Vision & Learning Laboratory
- Dual B.S. in Statistics and Artificial Intelligence, Seoul National University (Mar. 2017 – Feb. 2022)
- Exchange Student, Hong Kong University of Science and Technology (Fall 2019)
- Seoul Science High School (Mar. 2014 – Feb. 2017)
