Hi! I am Guozhong Li, currently a Postdoctoral Research Fellow at King Abdullah University of Science and Technology (KAUST), and I am very honored to be advised by Prof. Panos Kalnis. Prior to that, I was fortunate to complete my Ph.D. in Computer Science at Hong Kong Baptist University (HKBU), under the supervision of Prof. Byron Choi and Prof. Jianliang Xu. I was also a member of the HKBU Database Group.

🔍 Research

My research focuses on efficient and scalable learning, spanning modeling, indexing, compression, and reasoning over dynamic, high-dimensional data, aiming to capture long-range and multimodal dependencies while supporting downstream tasks such as classification, clustering, forecasting, retrieval, and error-bounded compression with both efficiency and effectiveness. I pursue this goal through advances in shape-based learning, efficient transformers, symbolic indexing, graph-driven representations, and unified frameworks for large-scale temporal reasoning.

To realize this vision, my work contributes to the following directions:

• Representation for Temporal Understanding
  Shape-based models and instance profile techniques are developed to enable scalable and explainable time series analysis.
  (ShapeNet: AAAI’21,Autoshape: ArXiv, IPS: ICDE’22, BSPCover: TKDE’22, leSAX: ICDE’25, Hierarchical TS Abstraction: KDD’26, [Learned Graph Structure for TSC - Ongoing], [Hierarchical TS Anomaly Detection- Ongoing])

• Transformer for Temporal Modeling
  Variable-position and data-driven attention mechanisms are designed to capture long-range dependencies in multivariate sequences.
  (SVP-T: AAAI’23, DARKER: PVLDB’24, [Symbolic Transformer for Time Series Forecasting - Ongoing], Time Series Tokens for Transformer - Ongoing, Efficient Multimodal Transformer - Ongoing, [Time Series Foundation Model for Forecasting - Ongoing], [Time Series Foundation Model for Classification - Ongoing], [Segmentation for TS Forecasting - Ongoing])

• Error-bounded Compression for Scientific/Spatialtemporal Data
  Leverage temporal graphs and LLMs to design frameworks that support controllable, error-bounded scientific data compression.
  (GraphComp: TKDE‘26, LLMComp: BigData’25, DNN4LSDC - Ongoing, [LSDComp - Ongoing], [SemanticComp - Ongoing])

• Efficient Inference and Coordination in LLMs
  System-level techniques to accelerate inference and enhance interaction among large language models, including KV cache optimization, attention runtime control, and multi-model collaboration.
  (CHESS: ArXiv, ART: Arxiv, [KV-cache ORCHestrator - Ongoing], [LAG - Ongoing], [Metirc Bench for KV Cache Pruning - Ongoing])

🔥 News

• 2026.05:  🎉 One paper HTSA has been accepted by KDD 2026!
• 2026.04:  🎉 One paper GraphComp has been accepted by TKDE!

🎓 Education

• Ph.D. in Computer Science, Hong Kong Baptist University (HKBU), Hong Kong SAR, China
  2017 – 2021
  Thesis: Shapelet Discovery for Time Series Analysis

• M.E. in Software Engineering, University of Electronic Science and Technology of China (UESTC), Chengdu, China
  2014 – 2017

• B.E. in Software Engineering, University of Electronic Science and Technology of China (UESTC), Chengdu, China
  2010 – 2014

🎖 Honors and Awards

• Research Performance Award, Hong Kong Baptist University, 2020, 2021.
• Excellent Teaching Assistant Performance Award, Hong Kong Baptist University, 2018, 2019.
• Teaching Assistant Performance Award, Hong Kong Baptist University, 2017, 2018, 2019.
• Outstanding Graduate Student, UESTC, 2014, 2017
• IBM-CSC Scholarship, IBM & Chinese Scholarship Council, 2015.
• National Scholarship, Ministry of Education, P.R.China. 2011, 2012, 2013,2016.