Skip to content

Latest commit

 

History

History
318 lines (292 loc) · 12 KB

REFERENCES.md

File metadata and controls

318 lines (292 loc) · 12 KB

Scientific Articles

We encourage you to also check out work by the group behind GluonTS. They are grouped according to topic and ordered chronographically.

GluonTS Overview

GluonTS: Probabilistic and Neural Time Series Modeling in Python

@article{maddix2019,
	Author = {Alexander Alexandrov, Konstantinos Benidis, Michael Bohlke-Schneider, Valentin Flunkert, Jan Gasthaus, Tim Januschowski, Danielle C. Maddix,  Syama Rangapuram, Davis Salinas, Jasper Schulz, Lorenzo Stella, Ali Caner T\" urkmen, Yuyang Wang},
	Journal = {Journal of Machine Learning Research},
	Title = {GluonTS: Probabilistic and Neural Time Series Modeling in Python},
	Year = {2020},
	Volume = {21},
	Number = {116},
	Pages = {1-6}
}

Methods

A number of the below methods are available in GluonTS.

Optimization methods for time series

@inproceedings{rangapuram2021,
	Author = {Yucheng Lu, Youngsuk Park, Lifan Chen, Yuyang Wang, Christopher De Sa, Dean Foster},
	Booktitle = {International Conference on Machine Learning},
	Title = {Variance Reduced Training with Stratified Sampling for Forecasting Models},
	Year = {2021}
}

Hierarchical forecasting models

@inproceedings{rangapuram2021,
	Author = {Syama S. Rangapuram, Lucien D Werner, Konstantinos Benidis, Pedro Mercado, Jan Gasthaus, Tim Januschowski},
	Booktitle = {International Conference on Machine Learning},
	Title = {End-to-End Learning of Coherent Probabilistic Forecasts for Hierarchical Time Series},
	Year = {2021}
}

Multi-variate forecasting models

Normalizing Kalman Filters

@inproceedings{bezene2020nkf,
	Author = {Emmanuel de B\'{e}zenac, Syama S. Rangapuram, Konstantinos Benidis, Michael Bohlke-Schneider, Richard Kurle, Lorenzo Stella, Hilaf Hasson, Patrick Gallinari, Tim Januschowski},
	Booktitle = {Advances in Neural Information Processing Systems},
	Title = {Normalizing Kalman Filters for Multivariate Time Series Analysis},
	Year = {2020}
}

A multivariate forecasting model

@inproceedings{salinas2019high,
	Author = {Salinas, David and Bohlke-Schneider, Michael and Callot, Laurent and Gasthaus, Jan},
	Booktitle = {Advances in Neural Information Processing Systems},
	Title = {High-Dimensional Multivariate Forecasting with Low-Rank Gaussian Copula Processes},
	Year = {2019}
}

Deep Probabilistic forecasting models

Particle Filters

@inproceedings{kurle20,
	Author = {Richard Kurle, Syama Rangapuram, Emmanuel de Bezenac, Stepuhan Günnemann, Jan Gasthaus},
	Booktitle = {Advances in Neural Information Processing Systems},
	Title = {Deep Rao-Blackwellised Particle Filters for Time Series Forecasting},
	Year = {2019}
}

Deep Factor models, a global-local forecasting method

@inproceedings{wang2019deep,
	Author = {Wang, Yuyang and Smola, Alex and Maddix, Danielle and Gasthaus, Jan and Foster, Dean and Januschowski, Tim},
	Booktitle = {International Conference on Machine Learning},
	Pages = {6607--6617},
	Title = {Deep factors for forecasting},
	Year = {2019}
}

DeepAR, an RNN-based probabilistic forecasting model

@article{flunkert2019deepar,
	Author = {Salinas, David and Flunkert, Valentin and Gasthaus, Jan and Tim Januschowski},
	Journal = {International Journal of Forecasting},
	Title = {DeepAR: Probabilistic forecasting with autoregressive recurrent networks},
	Year = {2019}
}

A flexible way to model probabilistic forecasts via spline quantile forecasts.

@inproceedings{gasthaus2019probabilistic,
	Author = {Gasthaus, Jan and Benidis, Konstantinos and Wang, Yuyang and Rangapuram, Syama Sundar and Salinas, David and Flunkert, Valentin and Januschowski, Tim},
	Booktitle = {The 22nd International Conference on Artificial Intelligence and Statistics},
	Pages = {1901--1910},
	Title = {Probabilistic Forecasting with Spline Quantile Function RNNs},
	Year = {2019}
}

Using RNNs to parametrize State Space Models.

@inproceedings{rangapuram2018deep,
	Author = {Rangapuram, Syama Sundar and Seeger, Matthias W and Gasthaus, Jan and Stella, Lorenzo and Wang, Yuyang and Januschowski, Tim},
	Booktitle = {Advances in Neural Information Processing Systems},
	Pages = {7785--7794},
	Title = {Deep state space models for time series forecasting},
	Year = {2018}
}

Intermittent Demand Forecasting with Renewal Processes

@inproceedings{turkmen2020idf,
	Author = {T\"{u}rkmen, Ali Caner and Januschowski, Tim and Wang, Yuyang and Cemgil, Ali Taylan},
	Booktitle = {arxiv},
	Title = {Intermittent Demand Forecasting with Renewal Processes},
	Year = {2020}
}

Using categorical distributions in forecasting

@inproceedings{rabanser2020discrete,
	Author = {Rabanser, Stephan and Januschowski, Tim and Salinas, David and Flunkert, Valentin and Gasthaus, Jan},
	Booktitle = {KDD Workshop on Mining and Learning From Time Series},
	Title = {The Effectiveness of Discretization in Forecasting: An Empirical Study on Neural Time Series Models},
	Year = {2020}
}

Anomaly detection models

Distributional Time Series Models for Anomaly Detection

@inproceedings{ayed20anomaly,
	Author = {Ayed, Fadhel and Stella, Lorenzo and Januschowski, Tim and Gasthaus, Jan},
	Booktitle = {AIOPs},
	Title = {Anomaly Detection at Scale: The Case for Deep Distributional Time Series Models},
	Year = {2020}
}

AutoODE

Physics-Based Time Series Models for Learning Dynamical Systems with Distribution Shifts

@inproceedings{wang2020,
	Author = {Wang, Rui and Maddix, Danielle and Faloutsos, Christos and Wang, Yuyang and Yu, Rose},
	Booktitle = {NeurIPS 2020 Machine Learning in Public Health (MLPH) Workshop},
	Title = {Bridging Physics-based and Data-driven modeling for Learning Dynamical Systems},
	Year = {2020}
}

Prior, related work

A scalable state space model. Note that code for this model is currently not available in GluonTS.

@inproceedings{seeger2016bayesian,
	Author = {Seeger, Matthias W and Salinas, David and Flunkert, Valentin},
	Booktitle = {Advances in Neural Information Processing Systems},
	Pages = {4646--4654},
	Title = {Bayesian intermittent demand forecasting for large inventories},
	Year = {2016}
}

Tutorials

Tutorials are available in bibtex and with accompanying material, in particular slides, linked from below.

WWW 2020

paper slides

@inproceedings{faloutsos2020forecasting,
  author = {Faloutsos, Christos and Flunkert, Valentin and Gasthaus, Jan and Januschowski, Tim and Wang, Yuyang},
  title = {Forecasting Big Time Series: Theory and Practice},
  year = {2020},
  booktitle = {Companion Proceedings of the Web Conference 2020},
  pages = {320–321},
  series = {WWW '20}
}

KDD 2019

paper slides

@inproceedings{faloutsos19forecasting,
  author    = {Faloutsos, Christos and
               Flunkert, Valentin and
               Gasthaus, Jan and
               Januschowski, Tim and
               Wang, Yuyang},
  title     = {Forecasting Big Time Series: Theory and Practice},
  booktitle = {Proceedings of the 25th {ACM} {SIGKDD} International Conference on
               Knowledge Discovery {\&} Data Mining, {KDD} 2019, Anchorage, AK,
               USA, August 4-8, 2019.},
  year      = {2019}
  }

SIGMOD 2019

paper supporting material

@inproceedings{faloutsos2019classical,
 author = {Faloutsos, Christos and Gasthaus, Jan and Januschowski, Tim and Wang, Yuyang},
 title = {Classical and Contemporary Approaches to Big Time Series Forecasting},
 booktitle = {Proceedings of the 2019 International Conference on Management of Data},
 series = {SIGMOD '19},
 publisher = {ACM},
 address = {New York, NY, USA},
 year = {2019}
} 

VLDB 2018

paper supporting material

@article{faloutsos2018forecasting,
	Author = {Faloutsos, Christos and Gasthaus, Jan and Januschowski, Tim and Wang, Yuyang},
	Journal = {Proceedings of the VLDB Endowment},
	Number = {12},
	Pages = {2102--2105},
	Title = {Forecasting big time series: old and new},
	Volume = {11},
	Year = {2018}
}

General audience

An overview of forecasting libraries in Python. paper to appear

@article{januschowski19open,
  title={Open-Source Forecasting Tools in Python},
  author={Januschowski, Tim and Gasthaus, Jan and Wang, Yuyang},
  journal={Foresight: The International Journal of Applied Forecasting},
  year={2019}
}

A commentary on the M4 competition and its classification of the participating methods into 'statistical' and 'ML' methods. The article proposes alternative criteria.

@article{januschowski19criteria,
title = {Criteria for classifying forecasting methods},
author = {Januschowski, Tim and Gasthaus, Jan and  Wang, Yuyang and Salinas, David and Flunkert, Valentin and Bohlke-Schneider, Michael and Callot, Laurent},
journal = {International Journal of Forecasting},
year = {2019}
}

The business forecasting problem landscape can be divided into strategic, tactical and operational forecasting problems.

@article{januschowski18a,
  title={A Classification of Business Forecasting Problems},
  author={Januschowski, Tim and Kolassa, Stephan},
  journal={Foresight: The International Journal of Applied Forecasting},
  year={2019},
  volume={52}, 
  pages={36-43}
}

A two-part article introducing deep learning for forecasting. part 2 part 1

@article{januschowski18deep2,
title = {Deep Learning for Forecasting: Current Trends and Challenges},
journal = {Foresight: The International Journal of Applied Forecasting},
year = {2018},
author = {Januschowski, Tim and Gasthaus, Jan and Wang, Yuyang and Rangapuram, Syama Sundar and Callot, Laurent},
volume = {51}, 
pages = {42-47}
}
@article{januschowski18deep,
  title = {Deep Learning for Forecasting},
  author = {Januschowski, Tim and Gasthaus, Jan and Wang, Yuyang and Rangapuram, Syama and Callot, Laurent},
  journal = {Foresight},
  year = {2018}
}

System Aspects

Resilient neural forecasting system.

@article{bohlke2020resilient,
	Author = {Bohlke-Schneider, Michael and Kapoor, Shubham and Januschowski, Tim},
	Journal = {DEEM'20: Proeccdings of the Fourth International Workshop on Data Management for End-to-End Machine Learning},
	Title = {Resilient Neural Forecasting Systems},
	Year = {2020}
}

A large-scale retail forecasting system.

@article{bose2017probabilistic,
	Author = {B{\"o}se, Joos-Hendrik and Flunkert, Valentin and Gasthaus, Jan and Januschowski, Tim and Lange, Dustin and Salinas, David and Schelter, Sebastian and Seeger, Matthias and Wang, Yuyang},
	Journal = {Proceedings of the VLDB Endowment},
	Number = {12},
	Pages = {1694--1705},
	Title = {Probabilistic demand forecasting at scale},
	Volume = {10},
	Year = {2017}
}