About me

I am a fourth-year Ph.D. student at the Computer Science Department, UMass Boston, USA, advised by Prof. Xiaohui Liang. I received my B.Eng. in Computer Science and Technology in 2020 from Nanjing Forestry University, China, advised by Prof. Ning Ye.

I’m broadly interested in deep learning and its applications. My recent research agenda centered around transfer learning under the large language model (LLM) era, with the aim of developing and customizing foundation models for various downstream tasks. I explored various topics under this agenda, including

• Reasoning with task relating (EMNLP 2024),
• Multi-modality (speech and vision language models) (INTERSPEECH 2021, Multi-modal KDD 2023)
• Targeted training data selection (INTERSPEECH 2022)
• Interpretability (ICASSP 2022)
• Privacy concerns (GLOBECOM 2024)

I also dive deep into speech-based dementia detection, a superhuman-level task that requires LLM to reason about dementia-related patterns that human experts may not know.

News

• 04/2025: Excited to share a new preprint: “Focus Directions Make Your Language Models Pay More Attention to Relevant Contexts.” [Paper]
• 01/2025: I will intern at fixie.ai this spring working on speech language models and speech generation.
• 11/2024: I gave a talk at Umass Boston CS 442/642 Cybersecurity in the Internet of Things about LLM prompt privacy protection. Check out the slide link.
• 10/2024: I will attend COLM 2024 in Philadelphia.
• 09/2024: “Adversarial Text Generation using Large Language Models for Dementia Detection” have been accepted to EMNLP 2024 (Main). Camera-ready version will be ready soon. Check paper using the link. See you in Miami!