My two main research interests are:

• Flow (ODE)-based generative neural networks.
• Conformal prediction for uncertainty quantification on data with dependency.

Please see Google Scholar for a complete list of works

---

---

➡️ Flow Neural networks

This slide is my summary of normalizing flow and flow matching methods, including their comparisons with diffusion models.

Published

• Flow-based Distributionally Robust Optimization, 2024 Arxiv JSAIT
  Chen Xu, Jonghyeok Lee, Xiuyuan Cheng, Yao Xie 
  – IEEE Journal on Selected Areas in Information Theory (JSAIT), 2024. 
  – Preliminary version presented at NeurIPS workshop M3L: Mathematics of Modern Machine Learning.
• Normalizing flow neural networks by JKO scheme, 2023 
  Arxiv
   NeurIPS 2023 Github Poster Slide
  Chen Xu, Xiuyuan Cheng, Yao Xie
  – Spotlight (3% of total 12343 submissions) in Part of Advances in Neural Information Processing Systems 37 (NeurIPS 2023).
• Invertible Neural Networks for Graph Prediction, 2022
  Arxiv JSAIT Github Poster  Slide 
  Chen Xu, Xiuyuan Cheng, Yao Xie
  – IEEE Journal on Selected Areas in Information Theory (JSAIT), 2022. Shorter paper available.

Preprints

• Local Flow Matching Generative Models, 2024 Arxiv
  Chen Xu, Xiuyuan Cheng, Yao Xie
• Computing high-dimensional optimal transport by flow neural networks, 2023 Arxiv
  Chen Xu, Xiuyuan Cheng, Yao Xie
  – Preliminary version presented at NeurIPS workshop OTML: Optimal Transport and Machine Learning.

---

---

➡️ Conformal prediction

Published

This slide is a summary of my current progress on time-series CP, where I gave a talk (Youtube recording) organized by Time Series Analysis And Forecasting Society (TAFS).

• Conformal prediction for multi-dimensional time-series Arxiv ICML 2024 Github  Poster Slide
  Chen Xu*, Hanyang Jiang*, Yao Xie
  *Equal contribution
  – Spotlight (3% of total 9653 submissions) in Proceedings of the 41st International Conference on Machine Learning, 2024 (ICML 2024).
  
• Sequential predictive conformal inference for time series, 2023.
  Arxiv ICML 2023 Github  Poster Slide
  Chen Xu, Yao Xie
  – Proceedings of the 40th International Conference on Machine Learning, 2023 (ICML 2023).
• Conformal prediction for time-series, 2023. Arxiv Github 
  Chen Xu, Yao Xie
  – Media coverage: (1) Demystifying EnbPI: Mastering Conformal Prediction Forecasting (2) Conformal Predictions in Time Series Forecasting.
  – Implementation: (1) MAPIE for scikit-learn (2) Fortuna from Amazon AWS with tutorial (3) Functime for time-series machine learning at scale (4) ConformalPrediction.jl in Julia (5) PUNCC in Python.
  – Journal version: IEEE Transactions on Pattern Analysis and Machine Intelligence (IF: 24.31).
  TPAMI Slide Poster 
  – Conference version: Oral paper (3% of total 5513 submissions) in the Proceedings of the 38th International Conference on Machine Learning, PMLR 139, 2021 (ICML 2021). ICML 2021
• Predictive inference is free with the jackknife+-after-bootstrap, 2020. NeurIPS 2020 Code
  Byol Kim, Chen Xu, Rina Barber
  – Part of Advances in Neural Information Processing Systems 34 (NeurIPS 2020).
• Conformal prediction set for time-series, 2022. Arxiv Poster Github
   
  Chen Xu, Yao Xie
  – Workshop on Distribution-Free Uncertainty Quantification at ICML 2022 (Strongly accepted).
  
• Conformal anomaly detection on spatio-temporal observations with missing data, 2021. Arxiv Poster Github 
  Chen Xu, Yao Xie
  – Workshop on Distribution-free Uncertainty Quantification at ICML 2021.

---

---

➡️ Miscellaneous

Published

• Spatio-Temporal Wildfire Prediction using Multi-Modal Data, 2023. Arxiv JSAIT Slide 
  Chen Xu, Daniel A. Zuniga Vazquez, Rui Yao, Feng Qiu, Yao Xie
  – IEEE Journal on Selected Areas in Information Theory (JSAIT), 2023. Shorter paper available.

Preprints

• Generalized generalized linear models: Convex estimation and online bounds, 2023 Arxiv 
  Anatoli Juditsky, Arkadi Nemirovski, Yao Xie, Chen Xu. (Authors listed alphabetically.)
  – Preprint
• Solar radiation ramping events modeling using spatio-temporal point processes, 2022. ArxivSlide 
  Minghe Zhang*, Chen Xu*, Andy Sun, Feng Qiu, Yao Xie
  *Equal contribution
  – Major Revision, INFORMS Journal on Data Science. Shorter paper available.
  – Earlier versions presented at (1) NSF AMPS 2020 (2) The 2021 INFORMS Conference on Service Science (ICSS 2021) (3) INFORMS 2021 General Session, Adaptive Online Learning of High-dimensional Data. 
• An alternative approach to train neural networks using monotone variational inequality, 2022 Arxiv
  Chen Xu, Xiuyuan Cheng, Yao Xie
  – Preliminary version presented at NeurIPS workshop OPT 2023: Optimization for Machine Learning.