Towards Utilitarian Online Learning – A Review of Online Algorithms in Open Feature Space
Surveys algorithms, applications, and open questions for online learning in open feature spaces.
I am an Assistant Professor in the Department of Data Science at William & Mary, within the School of Computing, Data Sciences, and Physics. Before joining William & Mary, I was an Assistant Professor in the Department of Computer Science at Old Dominion University from 2021 to 2024.
I received my Ph.D. from the Center for Advanced Computer Studies at the University of Louisiana at Lafayette in 2020, under the supervision of Prof. Xindong Wu, and my B.E. from Harbin Institute of Technology in 2013.
My research focuses on learning from evolving and networked data, with interests spanning data mining, streaming algorithms, graph learning, and AI applications in physical, environmental, and conservation sciences.
I am a recipient of the NSF CAREER Award (2026), NSF CRII Award (2023), and IEEE TCII Volunteer Award (2022). I am a Senior Member of the IEEE.
News
Archive-
Our paper “Physics-Refined Spatiotemporal Forecasting on Open-Boundary Hydrologic Graphs” has received the ICDM 2026 Best Paper Award.
-
Our team has received a $1.7 million NOAA award for research on AI-assisted sea scallop stock assessment. Kudos to lead PI Ming Sun (VIMS)!
-
I will serve as a Workshop Co-Chair for PAKDD 2027.
-
Our bridge proposal on streaming continual learning has been accepted for the AAAI-27 Bridge Program.
-
I have been appointed Director of Research Engagement & Outreach in the Department of Data Science at William & Mary.
-
I have been elevated to IEEE Senior Member.
-
Three papers on flow matching, graph learning from label proportions, and cold-start drug–drug interaction prediction have been accepted at NeurIPS 2026.
-
One paper on reinforcement-based augmentation for scientific sketch image retrieval has been accepted at IEEE BigData 2026.
-
One paper on AI-assisted analysis of surfclam shell growth bands has been accepted at the IEEE BigData 2026 High School Symposium. Congratulations to Sophia Sultan, our high school researcher from The Potomac School! I am glad to support K–12 students through their first research experiences.
-
Two papers on physics-informed hydrologic forecasting and architecture search for radiotherapy dose prediction have been accepted at IEEE ICDM 2026.
-
I have received an NSF CAREER Award. Thanks to NSF for supporting our research on graph dynamics and learning in changing environments.
Research & Selected Publications
Full listOnline Learning
Open and evolving feature spaces, delayed feedback, and changing data distributions.
Online Learning from Evolving Feature Spaces with Deep Variational Models
Transfers knowledge across evolving feature spaces through shared latent representations and adaptive model capacity.
Online Random Feature Forests for Learning in Varying Feature Spaces
Maintains interpretable decision forests as features appear or disappear in online data streams.
Online Learning for Wild Streaming Data with Delayed Feedback
Learns from heterogeneous, incomplete, and drifting streams when ground-truth feedback arrives late.
Graph Learning
Graph dynamics as entities and relationships evolve, alongside heterogeneous graph representations and ensemble learning.
Graph Learning from Label Proportions using Topology-Aware Pseudo Labeling and Aggregation
Learns graph neural networks from aggregate label proportions using topology-aware pseudo-labels and optimal transport.
Relink: Constructing Query-Driven Evidence Graph On-the-Fly for GraphRAG
Builds query-specific evidence graphs by recovering missing relations and selecting relevant facts for question answering.
HGEN: Heterogeneous Graph Ensemble Networks
Combines diverse heterogeneous-graph learners to balance predictive accuracy and representational diversity.
HGDL: Heterogeneous Graph Label Distribution Learning
Learns node label distributions while preserving structural and semantic consistency in heterogeneous graphs.
AI for Science and Conservation
Physics-informed prediction and scientific image analysis, with applications in environmental science, medicine, and marine conservation.
Physics-Refined Spatiotemporal Forecasting on Open-Boundary Hydrologic Graphs
Uses boundary proxies and physics refinements to stabilize forecasting on open-boundary hydrologic graphs.
Topology-aware Neural Flux Prediction Guided by Physics
Incorporates directional graph structure and physical constraints into neural flux prediction.
Interpretable Individual Manatee Identification Using Sketch Images and Deep Neural Networks
Uses distinctive markings in expert-drawn sketches for interpretable identification of individual manatees.
A Survey on Computational Pathology Foundation Models: Datasets, Adaptation Strategies, and Evaluation Tasks
Surveys pathology foundation models, their pretraining datasets, adaptation strategies, and evaluation tasks.