About Me

Welcome! I am a final-year Ph.D. student at the Department of Computer Science, Emory University, where I am fortunate to be advised by Dr. Liang Zhao. Previously, I received my master’s degree in Statistics from George Washington University in 2020. I received my bachelor’s degree in Mathematics from the School of Mathematical Science, Fudan University in Shanghai, China in 2018. I worked as a research intern at Argonne National Laboratory and NEC Lab America.

News: I am actively looking for Machine Learning Engineer / Applied Scientist / Research Scientist roles starting anytime in 2025. Feel free to DM me!

Research Interests

I am passionate about designing efficient and generalizable learning algorithms with applications across diverse domains. My research focuses on three primary areas:

1. Efficient Large-Scale Machine Learning: Developing methods to optimize large-scale models, particularly focusing on model compression, inference optimization, and distributed training for deep learning systems, including LLMs.
2. Domain and Knowledge Transfer: Enhancing machine learning models’ adaptability across various domains/tasks, including multi-task learning, domain adaptation, and domain generalization, with a particular focus on data with temporal concept drift.
3. Neuro-Inspired Continual Learning: Designing lifelong learning algorithms inspired by neuroscience, incorporating memory replay mechanisms for efficiency and robustness.

---

Selected Projects

1. Efficient Large-Scale Machine Learning

Exploring scalable and efficient solutions for machine learning systems, with a focus on LLM inference optimization.

a) Inference Optimization of LLMs

• SparseLLM: Towards Global Pruning for Pre-trained Language Models
  NeurIPS 2024
• FedSpaLLM: Federated Pruning of Large Language Models
  Preprint
• Beyond Efficiency: A Systematic Survey of Resource-Efficient Large Language Models
  Under review of CSUR

b) Distributed Training for Graph Neural Networks

• Staleness-Alleviated Distributed GNN Training via Online Dynamic-Embedding Prediction
  Preprint

---

2. Domain and Knowledge Transfer

Enhancing machine learning models’ adaptability and effectiveness across various domains and tasks, with a focus on temporal domain generalization and multi-task learning.

a) Domain Adaptation/Generalization

• Prompt-Based Domain Discrimination for Multi-source Time Series Domain Adaptation
  KDD 2024
• Continuous Temporal Domain Generalization
  NeurIPS 2024
• Temporal Domain Generalization with Drift-Aware Dynamic Neural Networks
  ICLR 2023 (Oral)

b) Multi-Task Learning

• Sign-Regularized Multi-Task Learning
  SDM 2023
• Saliency-Regularized Deep Multi-Task Learning
  KDD 2022

---

3. Neuro-Inspired Continual Learning

Developing lifelong learning approaches with efficient memory-replay mechanisms and neuro-inspiration.

• Saliency-Augmented Memory Completion for Continual Learning
  SDM 2023

Services and Awards

• PC member for KDD, ICML, ICLR, AISTATS, NeurIPS, AAAI, ICDM, etc.
• Primary writer for the NSF NAIRR 240189 grant ($15k) on parallel and distributed training of LLMs on graphs.
• KDD 22’, ICLR 23’, SDM 23’, CIKM 23’, NeurIPS 24’ student travel award