Tampere University of Technology

TUTCRIS Research Portal

Deep Adaptive Input Normalization for Time Series Forecasting

Research output: Contribution to journalArticleScientificpeer-review

Standard

Deep Adaptive Input Normalization for Time Series Forecasting. / Passalis, Nikolaos; Tefas, Anastasios; Kanniainen, Juho; Gabbouj, Moncef; Iosifidis, Alexandros.

In: IEEE Transactions on Neural Networks and Learning Systems , 18.12.2019.

Research output: Contribution to journalArticleScientificpeer-review

Harvard

Passalis, N, Tefas, A, Kanniainen, J, Gabbouj, M & Iosifidis, A 2019, 'Deep Adaptive Input Normalization for Time Series Forecasting', IEEE Transactions on Neural Networks and Learning Systems . https://doi.org/10.1109/TNNLS.2019.2944933

APA

Passalis, N., Tefas, A., Kanniainen, J., Gabbouj, M., & Iosifidis, A. (2019). Deep Adaptive Input Normalization for Time Series Forecasting. IEEE Transactions on Neural Networks and Learning Systems . https://doi.org/10.1109/TNNLS.2019.2944933

Vancouver

Passalis N, Tefas A, Kanniainen J, Gabbouj M, Iosifidis A. Deep Adaptive Input Normalization for Time Series Forecasting. IEEE Transactions on Neural Networks and Learning Systems . 2019 Dec 18. https://doi.org/10.1109/TNNLS.2019.2944933

Author

Passalis, Nikolaos ; Tefas, Anastasios ; Kanniainen, Juho ; Gabbouj, Moncef ; Iosifidis, Alexandros. / Deep Adaptive Input Normalization for Time Series Forecasting. In: IEEE Transactions on Neural Networks and Learning Systems . 2019.

Bibtex - Download

@article{d30c6b7a975247e997b67513d7fe4672,
title = "Deep Adaptive Input Normalization for Time Series Forecasting",
abstract = "Deep learning (DL) models can be used to tackle time series analysis tasks with great success. However, the performance of DL models can degenerate rapidly if the data are not appropriately normalized. This issue is even more apparent when DL is used for financial time series forecasting tasks, where the nonstationary and multimodal nature of the data pose significant challenges and severely affect the performance of DL models. In this brief, a simple, yet effective, neural layer that is capable of adaptively normalizing the input time series, while taking into account the distribution of the data, is proposed. The proposed layer is trained in an end-to-end fashion using backpropagation and leads to significant performance improvements compared to other evaluated normalization schemes. The proposed method differs from traditional normalization methods since it learns how to perform normalization for a given task instead of using a fixed normalization scheme. At the same time, it can be directly applied to any new time series without requiring retraining. The effectiveness of the proposed method is demonstrated using a large-scale limit order book data set, as well as a load forecasting data set.",
keywords = "Data normalization, deep learning (DL), limit order book data, time series forecasting.",
author = "Nikolaos Passalis and Anastasios Tefas and Juho Kanniainen and Moncef Gabbouj and Alexandros Iosifidis",
note = "EXT={"}Tefas, Anastasios{"} EXT={"}Iosifidis, Alexandros{"}",
year = "2019",
month = "12",
day = "18",
doi = "10.1109/TNNLS.2019.2944933",
language = "English",
journal = "IEEE Transactions on Neural Networks and Learning Systems",
issn = "2162-237X",
publisher = "IEEE Computational Intelligence Society",

}

RIS (suitable for import to EndNote) - Download

TY - JOUR

T1 - Deep Adaptive Input Normalization for Time Series Forecasting

AU - Passalis, Nikolaos

AU - Tefas, Anastasios

AU - Kanniainen, Juho

AU - Gabbouj, Moncef

AU - Iosifidis, Alexandros

N1 - EXT="Tefas, Anastasios" EXT="Iosifidis, Alexandros"

PY - 2019/12/18

Y1 - 2019/12/18

N2 - Deep learning (DL) models can be used to tackle time series analysis tasks with great success. However, the performance of DL models can degenerate rapidly if the data are not appropriately normalized. This issue is even more apparent when DL is used for financial time series forecasting tasks, where the nonstationary and multimodal nature of the data pose significant challenges and severely affect the performance of DL models. In this brief, a simple, yet effective, neural layer that is capable of adaptively normalizing the input time series, while taking into account the distribution of the data, is proposed. The proposed layer is trained in an end-to-end fashion using backpropagation and leads to significant performance improvements compared to other evaluated normalization schemes. The proposed method differs from traditional normalization methods since it learns how to perform normalization for a given task instead of using a fixed normalization scheme. At the same time, it can be directly applied to any new time series without requiring retraining. The effectiveness of the proposed method is demonstrated using a large-scale limit order book data set, as well as a load forecasting data set.

AB - Deep learning (DL) models can be used to tackle time series analysis tasks with great success. However, the performance of DL models can degenerate rapidly if the data are not appropriately normalized. This issue is even more apparent when DL is used for financial time series forecasting tasks, where the nonstationary and multimodal nature of the data pose significant challenges and severely affect the performance of DL models. In this brief, a simple, yet effective, neural layer that is capable of adaptively normalizing the input time series, while taking into account the distribution of the data, is proposed. The proposed layer is trained in an end-to-end fashion using backpropagation and leads to significant performance improvements compared to other evaluated normalization schemes. The proposed method differs from traditional normalization methods since it learns how to perform normalization for a given task instead of using a fixed normalization scheme. At the same time, it can be directly applied to any new time series without requiring retraining. The effectiveness of the proposed method is demonstrated using a large-scale limit order book data set, as well as a load forecasting data set.

KW - Data normalization

KW - deep learning (DL)

KW - limit order book data

KW - time series forecasting.

U2 - 10.1109/TNNLS.2019.2944933

DO - 10.1109/TNNLS.2019.2944933

M3 - Article

JO - IEEE Transactions on Neural Networks and Learning Systems

JF - IEEE Transactions on Neural Networks and Learning Systems

SN - 2162-237X

ER -