xtreme

xTreMe: Prediction and Characterization of Extreme Events in Spatio-Temporal Data

This repository contains various datasets and software implemented as part of a project funded by the National Science Foundation. The goal of this project is to develop novel algorithms for predicting and characterizing extreme events in large-scale spatio-temporal data. For more information, please visit the project website.

List of Software:

• SELF-RECOVER[8]: This repository contains the python implementation for forecasting extreme values from multiple time series containing missing values via a self-supervised learning approach.
• DEMM[7]: This repository contains the python implementation of a deep learning-based hurdle model with extreme value theory to enable point and distribution prediction of zero-inflated, heavy-tailed spatiotemporal variables.
• DeepExtrema[6]: This repository contains the python implementation of a deep neural network framework for forecasting the block maximum value of a time series according to the generalized extreme value (GEV) distribution.
• COMETFlows[5]: This repository contains the python implementation of a deep generative model architecture designed to capture asymmetric heavy-tailedness and tail dependence in multivariate data.
• DeepGPD[4]: This repository contains the python implementation of a deep learning framework for long-term prediction of the distribution of extreme values at different locations based on the generalized Pareto distribution.
• PRL[3]: This repository contains the official implementation of the PRL algorithm for learning deep neural networks under agnostic corrupted supervision.
• RCA[2]: This repository contains the official python implementation of the Robust Collaborative Autoencoders(RCA) algorithm for anomaly detection.
• JOHAN[1]: This repository contains the Matlab implementation of the JOHAN Algorithm for joint hurricane trajectory and intensity forecasting.

Publications:

1. Ding Wang and Pang-Ning Tan. JOHAN: A Joint Online Hurricane Trajectory and Intensity Forecasting Framework. In Proceedings of ACM SIGKDD International Conf on Data Mining (KDD 2021), 2021.
2. Boyang Liu, Ding Wang, Kaixiang Lin, Pang-Ning Tan, and Jiayu Zhou. RCA: A Deep Collaborative Autoencoder Approach for Anomaly Detection. In Proceedings of 30th International Joint Conference on Artificial Intelligence (IJCAI 2021), 2021.
3. Boyang Liu, Mengying Sun, Ding Wang, Pang-Ning Tan, and Jiayu Zhou. Learning Deep Neural Networks under Agnostic Corrupted Supervision. In Proceedings of International Conf on Machine Learning (ICML 2021), 2021.
4. Tyler Wilson, Pang-Ning Tan, and Lifeng Luo. DeepGPD: A Deep Learning Approach for Modeling Geospatio-Temporal Extreme Events. In Proceedings of the 36th AAAI Conference on Artificial Intelligence (AAAI 2022), 2022.
5. Andrew McDonald, Pang-Ning Tan, and Lifeng Luo COMET Flows: Towards Generative Modeling of Multivariate Extremes and Tail Dependence. In Proceedings of the 31st International Joint Conference on Artificial Intelligence (IJCAI 2022), Vienna, Austria, 2022.
6. Asadullah Hill Galib, Andrew McDonald, Tyler Wilson, Lifeng Luo, and Pang-Ning Tan. DeepExtrema: A Deep Learning Approach for Forecasting Block Maxima in Time Series Data. In Proceedings of the 31st International Joint Conference on Artificial Intelligence (IJCAI 2022), Vienna, Austria, 2022.
7. Tyler Wilson, Andrew McDonald, Asadullah Galib, Pang-Ning Tan, and Lifeng Luo. Beyond Point Prediction: Capturing Zero-Inflated & Heavy-Tailed Spatiotemporal Data with Deep Extreme Mixture Models. In Proceedings of ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD 2022), Washington, DC, 2022.
8. Asadullah Hill Galib, Andrew McDonald, Tyler Wilson, Lifeng Luo, and Pang-Ning Tan. Self-Recover: Forecasting Block Maxima in Time Series from Predictors with Disparate Temporal Coverage using Self-Supervised Learning. In Proceedings of the 32nd International Joint Conference on Artificial Intelligence (IJCAI 2023), Macao, 2023.
9. Asadullah Hill Galib, Lifeng Luo, and Pang-Ning Tan. SimEXT: Self-supervised Representation Learning for Extreme Values in Time Series. . To appear in Proc of the 23rd IEEE International Conference on Data Mining (ICDM 2023), Shanghai, 2023.

Acknowledgments:

This project is supported by the National Science Foundation under grant #IIS-2006633. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.