My research goal is to develop computationally and data-efficient machine learning models and algorithms for real-world applications, grounded in meta-learning and meta-modeling. I am particularly interested in extending these approaches to natural language processing, self-supervised learning, dataset condensation, hyperparameter optimization, AI safety, privacy, and multi-modal learning.
Education
Advised by Sung Ju Hwang
Advised by Sung Ju Hwang
Experiences
Advised by Lei Song and Jiang Bian.
- Multi-domain test-time scaling with reward models
- Online data selection
Publications
arXiv 2025
ACL 2025 Findings
ICLR 2025
ICML 2021
ICLR 2021 (spotlight)