Yizhuo (Matthew) Zhang

I am Yizhuo (Matthew) Zhang. I graduated from University of Washington and Tsinghua University. I am interested in LLM applications, especially on agents quality, LLM / VLM applications. I also had research experience in quantitative finance, time-series forecasting and human-computer interaction.

Email  /  Semantic Scholar  /  LinkedIn

Research

I'm interested in the following areas:

• Agents
• LLM/VLM Applications
• Reasoning
• Human Computer/AI Interaction
• Time-series Forecasting and Applications

Employing post-training alignment algorithms against rigid memorizing patterns to unlock generalizable learning of graphs using synthetic data.

Exploring the use of users' gaze data as implicit prompts to personalize and improve responses in text-based LLM applications.

Performance evaluation and improvement of LLM graph reasoning generalization using supervised fine-tuning and DPO.

Lightweight deep learning architecture merely based on simple MLP-based structures, inspired by a crucial fact that down-sampling time series often preserves the majority of its information.

Redirection techniques decouple tracked physical motion and virtual motion, allowing users to explore virtual environments with more flexibility.

Velocity guided amplification approach for head rotation in VR headsets, enabling the amplification factor to be dynamically changed according to head rotation velocity, and keep it within the comfort range

Miscellanea