Yinghao Li

I’m a Ph.D. student majored in Machine Learning @ Georgia Institute of Technology, advised by Dr. Chao Zhang and Dr. Le Song.

My research interest covers

Information Extraction
- Effective and efficient way of fine-tuning and using large language models to retrieve structured information from free-formed documents;
- Using LLMs as pseudo dataset generator to supervise the training of smaller task-specific labelers, such as BERT;
- Weakly-supervised named entity recognision and text classificatio with multiple noisy labeling functions [Conditional Hidden Markov Model (CHMM). 2021, Sparse CHMM. 2022, Wrench. 2021, Ren et al. 2020];
- HTML information extraction through Transformer-based DOM node classifier [TrENC. 2023];
- Rule-based molecular property extractor [post, Toland et al. 2023].
LLM Reasoning
- Investigating the source of reasoning ability of LLM: is it intrinsic or a simple mimic of training data? [Minesweeper. 2023].
Uncertainty Estimation
- Benchmarking the uncertainty quantification methods for large modecular representation models on molecular property prediction [MUBen].
Text Generation
- Generate paraphrases under the syntactic guidance of constituency parsing tags [Li et al. 2020].

I got my Master degree from Georgia Tech, School of ECE, where I worked with Prof. Ying Zhang on Radar SCG signal processing and understanding [Li et al. 2020, Xia et al. 2021].

Education

Ph.D. student @ Georgia Institute of Technology Machine Learning, School of Electrical and Computer Engineering

Master of Science @ Georgia Institute of Technology School of Electrical and Computer Engineering

Bachelor of Science @ Southeast University, Nanjing School of Instrument Science and Engineering

Applied scientist intern @ AWS, New York SAAR

Applied scientist intern @ Amazon, Seattle Product Graph