Publications
Research in temporal and graph learning, AI for health, and spatiotemporal modeling.
2026
Temporally Detailed Hypergraph Neural ODE for Disease Progression Modeling
Tingsong Xiao, Yao An Lee, Zelin Xu, Yupu Zhang, Zibo Liu, Yu Huang, Jiang Bian, Jingchuan Guo, Zhe Jiang
The 14th International Conference on Learning Representations (ICLR 2026)
We propose Temporally Detailed Hypergraph Neural Ordinary Differential Equation (TD-HNODE), which represents disease progression on clinically recognized trajectories as a temporally detailed hypergraph and learns the continuous-time progression dynamics via a neural ODE framework.
Temporally Detailed Hypergraph Neural ODE for Disease Progression Modeling
Tingsong Xiao, Yao An Lee, Zelin Xu, Yupu Zhang, Zibo Liu, Yu Huang, Jiang Bian, Jingchuan Guo, Zhe Jiang
The 14th International Conference on Learning Representations (ICLR 2026)
We propose Temporally Detailed Hypergraph Neural Ordinary Differential Equation (TD-HNODE), which represents disease progression on clinically recognized trajectories as a temporally detailed hypergraph and learns the continuous-time progression dynamics via a neural ODE framework.
A Survey on Uncertainty Quantification Methods for Deep Learning
Wenchong He, Zhe Jiang, Tingsong Xiao, Zelin Xu, Yukun Li
ACM Computing Surveys (CSUR 2026)
A survey of uncertainty quantification in deep learning, organizing methods by the sources of uncertainty they address in models and data, and discussing applications and open research directions.
A Survey on Uncertainty Quantification Methods for Deep Learning
Wenchong He, Zhe Jiang, Tingsong Xiao, Zelin Xu, Yukun Li
ACM Computing Surveys (CSUR 2026)
A survey of uncertainty quantification in deep learning, organizing methods by the sources of uncertainty they address in models and data, and discussing applications and open research directions.
2025
DecoyDB: A Dataset for Graph Contrastive Learning in Protein-Ligand Binding Affinity Prediction
Yupu Zhang, Zelin Xu, Tingsong Xiao, Gustavo Seabra, Yanjun Li, Chenglong Li, Zhe Jiang
The 39th Annual Conference on Neural Information Processing Systems (NeurIPS 2025)
We propose DecoyDB, a large-scale, structure-aware dataset for self-supervised graph contrastive learning on protein–ligand complexes. DecoyDB consists of high-resolution ground truth complexes and diverse decoy structures with RMSD annotations. We design a customized GCL framework to pretrain graph neural networks and demonstrate superior accuracy, sample efficiency, and generalizability.
DecoyDB: A Dataset for Graph Contrastive Learning in Protein-Ligand Binding Affinity Prediction
Yupu Zhang, Zelin Xu, Tingsong Xiao, Gustavo Seabra, Yanjun Li, Chenglong Li, Zhe Jiang
The 39th Annual Conference on Neural Information Processing Systems (NeurIPS 2025)
We propose DecoyDB, a large-scale, structure-aware dataset for self-supervised graph contrastive learning on protein–ligand complexes. DecoyDB consists of high-resolution ground truth complexes and diverse decoy structures with RMSD annotations. We design a customized GCL framework to pretrain graph neural networks and demonstrate superior accuracy, sample efficiency, and generalizability.
Spatio-temporal Partial Sensing Forecast of Long-term Traffic
Zibo Liu, Zhe Jiang, Zelin Xu, Tingsong Xiao, Zhengkun Xiao, Yupu Zhang, Haibo Wang, Shigang Chen
Transactions on Machine Learning Research (TMLR 2025)
SLPF forecasts long-term traffic at locations without sensors. It combines rank-based embeddings, spatial transfer, and multi-step training to address noisy observations and the shift between sensed and unsensed locations.
Spatio-temporal Partial Sensing Forecast of Long-term Traffic
Zibo Liu, Zhe Jiang, Zelin Xu, Tingsong Xiao, Zhengkun Xiao, Yupu Zhang, Haibo Wang, Shigang Chen
Transactions on Machine Learning Research (TMLR 2025)
SLPF forecasts long-term traffic at locations without sensors. It combines rank-based embeddings, spatial transfer, and multi-step training to address noisy observations and the shift between sensed and unsensed locations.
CoastalBench: A Decade-Long High-Resolution Dataset to Emulate Complex Coastal Processes
Zelin Xu, Yupu Zhang, Tingsong Xiao, Maitane Olabarrieta Lizaso, Jose M Gonzalez-Ondina, Zibo Liu, Shigang Chen, Zhe Jiang
The 42nd International Conference on Machine Learning (ICML 2025)
We introduce a decade-long, high-resolution (<100m) coastal circulation modeling dataset in southwest Florida with ~6 million cells. The dataset contains key oceanography variables and external forcings. We evaluate a customized Vision Transformer model for predicting ocean variables at varying lead times, providing a benchmark for deep learning models in high-resolution coastal simulations.
CoastalBench: A Decade-Long High-Resolution Dataset to Emulate Complex Coastal Processes
Zelin Xu, Yupu Zhang, Tingsong Xiao, Maitane Olabarrieta Lizaso, Jose M Gonzalez-Ondina, Zibo Liu, Shigang Chen, Zhe Jiang
The 42nd International Conference on Machine Learning (ICML 2025)
We introduce a decade-long, high-resolution (<100m) coastal circulation modeling dataset in southwest Florida with ~6 million cells. The dataset contains key oceanography variables and external forcings. We evaluate a customized Vision Transformer model for predicting ocean variables at varying lead times, providing a benchmark for deep learning models in high-resolution coastal simulations.
XTSFormer: Cross-Temporal-Scale Transformer for Irregular-Time Event Prediction in Clinical Applications
Tingsong Xiao, Zelin Xu, Wenchong He, Zhengkun Xiao, Yupu Zhang, Zibo Liu, Shigang Chen, My T. Thai, Jiang Bian, Parisa Rashidi, Zhe Jiang
The 39th Annual AAAI Conference on Artificial Intelligence (AAAI 2025)
Modeling irregularly timed clinical events from EHRs is challenging due to time irregularity, cyclical patterns, and multi-scale event interactions. We propose XTSFormer, a transformer-based model with cycle-aware time encoding and multi-scale attention to address these challenges.
XTSFormer: Cross-Temporal-Scale Transformer for Irregular-Time Event Prediction in Clinical Applications
Tingsong Xiao, Zelin Xu, Wenchong He, Zhengkun Xiao, Yupu Zhang, Zibo Liu, Shigang Chen, My T. Thai, Jiang Bian, Parisa Rashidi, Zhe Jiang
The 39th Annual AAAI Conference on Artificial Intelligence (AAAI 2025)
Modeling irregularly timed clinical events from EHRs is challenging due to time irregularity, cyclical patterns, and multi-scale event interactions. We propose XTSFormer, a transformer-based model with cycle-aware time encoding and multi-scale attention to address these challenges.
2024
Spatial-Logic-Aware Weakly Supervised Learning for Flood Mapping on Earth Imagery
Zelin Xu, Tingsong Xiao, Wenchong He, Yu Wang, Zhe Jiang, Shigang Chen, Yiqun Xie, Xiaowei Jia, Da Yan, Yang Zhou
The 38th Annual AAAI Conference on Artificial Intelligence (AAAI 2024)
This work addresses the challenge of flood mapping on high-resolution Earth imagery with coarse, noisy labels by proposing a spatial-logic-aware weakly supervised learning framework. The framework integrates symbolic spatial logic inference with probabilistic learning and introduces a multi-resolution spatial reasoning algorithm to handle high-resolution imagery efficiently. Experiments on real-world flood datasets demonstrate superior prediction accuracy compared to traditional and neural-symbolic baselines.
Spatial-Logic-Aware Weakly Supervised Learning for Flood Mapping on Earth Imagery
Zelin Xu, Tingsong Xiao, Wenchong He, Yu Wang, Zhe Jiang, Shigang Chen, Yiqun Xie, Xiaowei Jia, Da Yan, Yang Zhou
The 38th Annual AAAI Conference on Artificial Intelligence (AAAI 2024)
This work addresses the challenge of flood mapping on high-resolution Earth imagery with coarse, noisy labels by proposing a spatial-logic-aware weakly supervised learning framework. The framework integrates symbolic spatial logic inference with probabilistic learning and introduces a multi-resolution spatial reasoning algorithm to handle high-resolution imagery efficiently. Experiments on real-world flood datasets demonstrate superior prediction accuracy compared to traditional and neural-symbolic baselines.
2023
A Hierarchical Spatial Transformer for Massive Point Samples in Continuous Space
Wenchong He, Zhe Jiang, Tingsong Xiao, Zelin Xu, Shigang Chen, Ronald Fick, Miles Medina, Christine Angelini
The 37th Annual Conference on Neural Information Processing Systems (NeurIPS 2023)
This paper proposes a hierarchical spatial transformer for massive point samples, addressing challenges like long-range dependencies, irregular distributions, and high computational costs. Using quad-tree-based multi-resolution learning, efficient spatial attention, and uncertainty quantification, the model outperforms baselines and scales to one million points on a single NVIDIA A100 GPU.
A Hierarchical Spatial Transformer for Massive Point Samples in Continuous Space
Wenchong He, Zhe Jiang, Tingsong Xiao, Zelin Xu, Shigang Chen, Ronald Fick, Miles Medina, Christine Angelini
The 37th Annual Conference on Neural Information Processing Systems (NeurIPS 2023)
This paper proposes a hierarchical spatial transformer for massive point samples, addressing challenges like long-range dependencies, irregular distributions, and high computational costs. Using quad-tree-based multi-resolution learning, efficient spatial attention, and uncertainty quantification, the model outperforms baselines and scales to one million points on a single NVIDIA A100 GPU.
M-GCN: Multi-scale Graph Convolutional Network for 3D Point Cloud Classification
Jing Hu, Xincheng Wang, Ziheng Liao, Tingsong Xiao# (# corresponding author)
IEEE International Conference on Multimedia and Expo (ICME 2023)
This paper introduces a Multi-Scale Graph Convolutional Network (M-GCN) to enhance 3D point cloud analysis by leveraging multi-scale feature fusion for richer local topological representations. By extracting geometric features across scales, M-GCN improves the representation power of point clouds. Experiments on ModelNet40 demonstrate state-of-the-art performance in 3D point cloud classification.
M-GCN: Multi-scale Graph Convolutional Network for 3D Point Cloud Classification
Jing Hu, Xincheng Wang, Ziheng Liao, Tingsong Xiao# (# corresponding author)
IEEE International Conference on Multimedia and Expo (ICME 2023)
This paper introduces a Multi-Scale Graph Convolutional Network (M-GCN) to enhance 3D point cloud analysis by leveraging multi-scale feature fusion for richer local topological representations. By extracting geometric features across scales, M-GCN improves the representation power of point clouds. Experiments on ModelNet40 demonstrate state-of-the-art performance in 3D point cloud classification.
Spatial Knowledge-Infused Hierarchical Learning: An Application in Flood Mapping on Earth Imagery
Zelin Xu, Tingsong Xiao, Wenchong He, Yu Wang, Zhe Jiang
The 31st ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL 2023) Best Paper Award
This paper proposes a Spatial Knowledge-Infused Hierarchical Learning (SKI-HL) framework for Earth imagery with limited training labels, addressing challenges like sparse input labels, spatial uncertainty, and high computational costs. SKI-HL iteratively infers labels within a multi-resolution hierarchy using uncertainty-aware modules for selective label inference and neural network training. Experiments on flood mapping datasets demonstrate superior performance compared to baseline methods.
Spatial Knowledge-Infused Hierarchical Learning: An Application in Flood Mapping on Earth Imagery
Zelin Xu, Tingsong Xiao, Wenchong He, Yu Wang, Zhe Jiang
The 31st ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL 2023) Best Paper Award
This paper proposes a Spatial Knowledge-Infused Hierarchical Learning (SKI-HL) framework for Earth imagery with limited training labels, addressing challenges like sparse input labels, spatial uncertainty, and high computational costs. SKI-HL iteratively infers labels within a multi-resolution hierarchy using uncertainty-aware modules for selective label inference and neural network training. Experiments on flood mapping datasets demonstrate superior performance compared to baseline methods.
2022
Dual-Graph Learning Convolutional Networks for Interpretable Alzheimer’s Disease Diagnosis
Tingsong Xiao, Lu Zeng, Xiaoshuang Shi, Xiaofeng Zhu, Guorong Wu
The 25th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2022) Early Accept & Oral
This paper proposes a dual-graph learning convolutional network (dGLCN) for interpretable Alzheimer’s disease diagnosis by jointly learning subject and feature graphs within a GCN framework. By iteratively updating these graphs, dGLCN enhances interpretability in both subjects and brain regions while improving generalizability despite limited or noisy data. Experiments on ADNI datasets demonstrate that dGLCN outperforms comparison methods in binary classification tasks.
Dual-Graph Learning Convolutional Networks for Interpretable Alzheimer’s Disease Diagnosis
Tingsong Xiao, Lu Zeng, Xiaoshuang Shi, Xiaofeng Zhu, Guorong Wu
The 25th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2022) Early Accept & Oral
This paper proposes a dual-graph learning convolutional network (dGLCN) for interpretable Alzheimer’s disease diagnosis by jointly learning subject and feature graphs within a GCN framework. By iteratively updating these graphs, dGLCN enhances interpretability in both subjects and brain regions while improving generalizability despite limited or noisy data. Experiments on ADNI datasets demonstrate that dGLCN outperforms comparison methods in binary classification tasks.