Recurrent Neural Networks for Time Series Forecasting

A special issue of Forecasting (ISSN 2571-9394). This special issue belongs to the section "Forecasting in Computer Science".

Deadline for manuscript submissions: closed (1 January 2023) | Viewed by 6035

Special Issue Editors


E-Mail Website
Guest Editor
Nuclear Engineering and Radiological Sciences, University of Michigan, 2355 Bonisteel Blvd, Ann Arbor, MI 48109, USA
Interests: machine learning; time series forecasting; uncertainty quantification; optimization; surrogate modeling; multiphysics simulations; nuclear power engineering; autonomous control

E-Mail Website
Guest Editor
Nuclear Energy and Fuel Cycle Division, Oak Ridge National Laboratory, PO Box 2008, Oak Ridge, TN 37831, USA
Interests: artificial intelligence; machine learning; time series forecasting; decision science; digital engineering; clean energy; nuclear engineering; modeling and simulation
Nuclear Science and Technology Division, Idaho National Laboratory, 1955 N. Fremont Avenue, Idaho Falls, ID 83415, USA
Interests: machine Learning; uncertainty quantification; time series forecasting; statistical inference; reduced-order modeling; nuclear engineering

E-Mail Website
Guest Editor
Modeling and Simulation Specialist, Department of Digital Reactor Technology and Development, Idaho National Laboratory, 1955 N Fremont Ave, Idaho Falls, ID 83415, USA
Interests: machine learning; deep learning; time series analysis; digital twins; reduced order modeling; uncertainty quantification; sensitivity analysis; metaheuristic optimization; transformer models; attention mechanisms

Special Issue Information

Dear Colleagues,

Recurrent neural networks (RNN) are a special family of neural networks that are designed for sequential data such as time series forecasting or natural language processing. RNN architectures are one of the major deep learning algorithms that are able to perform temporal analysis due to their internal memory, which stores information from the past. RNNs have seen several advancements, such as long short-term memory (LSTM), gated recurrent units (GRU), and convolutional LSTM (ConvLSTM), among others. Nevertheless, RNNs are still facing challenges in training costs, generalizability, performance with noisy data, and architecture optimization. This Special Issue invites researchers to present high-quality studies on RNN techniques for time series forecasting, ranging from theoretical studies to the development of innovative applications. Potential topics of interest include but are not limited to:

  • Novel algorithms exploring advancements in LSTM and GRU for time series forecasting.
  • Novel algorithms that combine RNN with advancements in machine learning such as bi-directional algorithms, attention mechanisms, transformers, and multiplicative LSTM for time series forecasting.
  • Novel algorithms that explore ConvLSTM architectures for spatio-temporal forecasting.
  • Novel algorithms allowing the use of evolutionary computation to optimize RNN weights and hyperparameters in forecasting.
  • Novel algorithms for time series forecasting under uncertainty with the use of Gaussian processes or Bayesian techniques.
  • Novel algorithms for time series forecasting with limited data.
  • Parallel and distributed RNN algorithms to accelerate training on GPUs.
  • Exploration of RNN autoencoders and/or transformers for time series anomaly detection.
  • Engineering applications of RNN algorithms in all disciplines are welcomed, while applications in carbon-free energy are particularly encouraged.

Dr. Majdi I. Radaideh
Dr. Xingang Zhao
Dr. Yifeng Che
Dr. Mohammad G. Abdo
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Forecasting is an international peer-reviewed open access quarterly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 1800 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

Keywords

  • recurrent neural networks
  • long short-term memory (LSTM)
  • gated recurrent units (GRU)
  • ConvLSTM
  • autoencoders
  • fault prognostics
  • anomaly detection
  • time series prediction with uncertainty
  • multivariate forecasting
  • signal processing
  • forecasting in energy

Published Papers (1 paper)

Order results
Result details
Select all
Export citation of selected articles as:

Research

21 pages, 1026 KiB  
Article
Time Series Dataset Survey for Forecasting with Deep Learning
by Yannik Hahn, Tristan Langer, Richard Meyes and Tobias Meisen
Forecasting 2023, 5(1), 315-335; https://doi.org/10.3390/forecast5010017 - 03 Mar 2023
Cited by 2 | Viewed by 5183
Abstract
Deep learning models have revolutionized research fields like computer vision and natural language processing by outperforming traditional models in multiple tasks. However, the field of time series analysis, especially time series forecasting, has not seen a similar revolution, despite forecasting being one of [...] Read more.
Deep learning models have revolutionized research fields like computer vision and natural language processing by outperforming traditional models in multiple tasks. However, the field of time series analysis, especially time series forecasting, has not seen a similar revolution, despite forecasting being one of the most prominent tasks of predictive data analytics. One crucial problem for time series forecasting is the lack of large, domain-independent benchmark datasets and a competitive research environment, e.g., annual large-scale challenges, that would spur the development of new models, as was the case for CV and NLP. Furthermore, the focus of time series forecasting research is primarily domain-driven, resulting in many highly individual and domain-specific datasets. Consequently, the progress in the entire field is slowed down due to a lack of comparability across models trained on a single benchmark dataset and on a variety of different forecasting challenges. In this paper, we first explore this problem in more detail and derive the need for a comprehensive, domain-unspecific overview of the state-of-the-art of commonly used datasets for prediction tasks. In doing so, we provide an overview of these datasets and improve comparability in time series forecasting by introducing a method to find similar datasets which can be utilized to test a newly developed model. Ultimately, our survey paves the way towards developing a single widely used and accepted benchmark dataset for time series data, built on the various frequently used datasets surveyed in this paper. Full article
(This article belongs to the Special Issue Recurrent Neural Networks for Time Series Forecasting)
Show Figures

Figure 1

Back to TopTop