Scholar
Stephen H. Bach
Google Scholar ID: hs6pGXoAAAAJ
Assistant Professor of Computer Science, Brown University
Citations & Impact
Citations
9,171
H-index
25
i10-index
35
Publications
20
Co-authors
8
list available
Publications
Resume (English only)
Academic Achievements
- Published multiple papers and received awards, including but not limited to:
- - Open-source release of the Bonito model
- - New pre-prints on generating training data for low-resource languages
- - New pre-prints on interpreting CLIP by learning to prompt it
- - Paper on follow-up prompting accepted to ICLR 2024
- - Paper on probing the compositional capabilities of CLIP accepted to EACL Findings 2024
- - Work on GPT-4 and low-resource languages won the Best Paper Award at NeurIPS Workshop on Socially Responsible Language Modelling Research (SoLaR) 2023
- - Paper exploring strategies for using CLIP as a pseudolabeler for prompt tuning will appear at NeurIPS 2023
- - Work on integrating large language models into weak supervision accepted to the ACM/IMS Journal of Data Science
Research Experience
- Leads the BATS machine learning research group. BATS stands for 'Bach's Awesome Team of Students'.
Education
- Eliot Horowitz Assistant Professor, Computer Science Department, Brown University, specific degree and advisor information not provided.
Background
- Research Interests: Improving the processes by which humans teach and instruct computers, including generalizing from fewer examples (like zero-shot and few-shot learning) and generating training data (like synthetic data generation and programmatic weak supervision). The goal is to reduce the effort required by people, especially non-computer scientists in specialized, technical domains, to get computers to do what they want. Application areas include information extraction, image understanding, scientific discovery, and other areas of data science.
