An autoencoder wavelet based deep neural network with attention mechanism for multi-step prediction of plant growth

Bashar Alhnaity; Stefanos Kollias; Georgios Leontidis; Shouyong Jiang; Bert Schamp; Simon Pearson

doi:10.1016/j.ins.2021.01.037

An autoencoder wavelet based deep neural network with attention mechanism for multi-step prediction of plant growth

Bashar Alhnaity^*, Stefanos Kollias, Georgios Leontidis, Shouyong Jiang, Bert Schamp, Simon Pearson

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

34 Citations (Scopus)

11 Downloads (Pure)

Abstract

Multi-step-ahead prediction is considered of major significance for time series analysis in many real life problems. Existing methods mainly focus on one-step-ahead forecasting, since multiple step forecasting generally fails due to accumulation of prediction errors. This paper presents a novel approach for predicting plant growth in agriculture, focusing on prediction of plant Stem Diameter Variations (SDV). The proposed approach consists of three main steps. At first, wavelet decomposition is applied to the original data, so as to facilitate model fitting and reduce noise. Then an encoder-decoder framework is developed using Long Short Term Memory (LSTM) and used for appropriate feature extraction from the data. Finally, a recurrent neural network including LSTM and an attention mechanism is proposed for modelling long-term dependencies in the time series data. Experimental results are presented which illustrate the good performance of the proposed approach and that it significantly outperforms the existing models, in terms of error criteria such as RMSE, MAE and MAPE.

Original language	English
Article number	INS16166
Pages (from-to)	35-50
Number of pages	16
Journal	Information Sciences
Volume	560
Early online date	1 Feb 2021
DOIs	https://doi.org/10.1016/j.ins.2021.01.037
Publication status	Published - 1 Jun 2021

Bibliographical note

Acknowledgements
This research was supported as part of SMARTGREEN, an Interreg project supported by the North Sea Programme of the European Regional Development Fund of the European Union. We would like to thank all growers (UK and EU), for providing us with the presented data sets. We also wish to thank the reviewers of the paper. Their valuable feedback, suggestions and comments helped us to improve the quality of this work.

Keywords

Multi-step prediction
Wavelet analysis
Deep neural networks
Attention mechanism
Time-series analysis
LSTM
Plant growth prediction

Access to Document

10.1016/j.ins.2021.01.037Licence: Unspecified

Alhnaity_etal_forecasting_plant_AAM
This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/.
Accepted author manuscript, 901 KBLicence: CC BY-NC-ND

Cite this

@article{139205663aa14b0bbc628f7a670c6f7f,

title = "An autoencoder wavelet based deep neural network with attention mechanism for multi-step prediction of plant growth",

abstract = "Multi-step-ahead prediction is considered of major significance for time series analysis in many real life problems. Existing methods mainly focus on one-step-ahead forecasting, since multiple step forecasting generally fails due to accumulation of prediction errors. This paper presents a novel approach for predicting plant growth in agriculture, focusing on prediction of plant Stem Diameter Variations (SDV). The proposed approach consists of three main steps. At first, wavelet decomposition is applied to the original data, so as to facilitate model fitting and reduce noise. Then an encoder-decoder framework is developed using Long Short Term Memory (LSTM) and used for appropriate feature extraction from the data. Finally, a recurrent neural network including LSTM and an attention mechanism is proposed for modelling long-term dependencies in the time series data. Experimental results are presented which illustrate the good performance of the proposed approach and that it significantly outperforms the existing models, in terms of error criteria such as RMSE, MAE and MAPE.",

keywords = "Multi-step prediction, Wavelet analysis, Deep neural networks, Attention mechanism, Time-series analysis, LSTM, Plant growth prediction",

author = "Bashar Alhnaity and Stefanos Kollias and Georgios Leontidis and Shouyong Jiang and Bert Schamp and Simon Pearson",

note = "Acknowledgements This research was supported as part of SMARTGREEN, an Interreg project supported by the North Sea Programme of the European Regional Development Fund of the European Union. We would like to thank all growers (UK and EU), for providing us with the presented data sets. We also wish to thank the reviewers of the paper. Their valuable feedback, suggestions and comments helped us to improve the quality of this work.",

year = "2021",

month = jun,

day = "1",

doi = "10.1016/j.ins.2021.01.037",

language = "English",

volume = "560",

pages = "35--50",

journal = "Information Sciences",

issn = "0020-0255",

publisher = "Elsevier Inc.",

}

TY - JOUR

T1 - An autoencoder wavelet based deep neural network with attention mechanism for multi-step prediction of plant growth

AU - Alhnaity, Bashar

AU - Kollias, Stefanos

AU - Leontidis, Georgios

AU - Jiang, Shouyong

AU - Schamp, Bert

AU - Pearson, Simon

N1 - Acknowledgements This research was supported as part of SMARTGREEN, an Interreg project supported by the North Sea Programme of the European Regional Development Fund of the European Union. We would like to thank all growers (UK and EU), for providing us with the presented data sets. We also wish to thank the reviewers of the paper. Their valuable feedback, suggestions and comments helped us to improve the quality of this work.

PY - 2021/6/1

Y1 - 2021/6/1

N2 - Multi-step-ahead prediction is considered of major significance for time series analysis in many real life problems. Existing methods mainly focus on one-step-ahead forecasting, since multiple step forecasting generally fails due to accumulation of prediction errors. This paper presents a novel approach for predicting plant growth in agriculture, focusing on prediction of plant Stem Diameter Variations (SDV). The proposed approach consists of three main steps. At first, wavelet decomposition is applied to the original data, so as to facilitate model fitting and reduce noise. Then an encoder-decoder framework is developed using Long Short Term Memory (LSTM) and used for appropriate feature extraction from the data. Finally, a recurrent neural network including LSTM and an attention mechanism is proposed for modelling long-term dependencies in the time series data. Experimental results are presented which illustrate the good performance of the proposed approach and that it significantly outperforms the existing models, in terms of error criteria such as RMSE, MAE and MAPE.

AB - Multi-step-ahead prediction is considered of major significance for time series analysis in many real life problems. Existing methods mainly focus on one-step-ahead forecasting, since multiple step forecasting generally fails due to accumulation of prediction errors. This paper presents a novel approach for predicting plant growth in agriculture, focusing on prediction of plant Stem Diameter Variations (SDV). The proposed approach consists of three main steps. At first, wavelet decomposition is applied to the original data, so as to facilitate model fitting and reduce noise. Then an encoder-decoder framework is developed using Long Short Term Memory (LSTM) and used for appropriate feature extraction from the data. Finally, a recurrent neural network including LSTM and an attention mechanism is proposed for modelling long-term dependencies in the time series data. Experimental results are presented which illustrate the good performance of the proposed approach and that it significantly outperforms the existing models, in terms of error criteria such as RMSE, MAE and MAPE.

KW - Multi-step prediction

KW - Wavelet analysis

KW - Deep neural networks

KW - Attention mechanism

KW - Time-series analysis

KW - LSTM

KW - Plant growth prediction

U2 - 10.1016/j.ins.2021.01.037

DO - 10.1016/j.ins.2021.01.037

M3 - Article

SN - 0020-0255

VL - 560

SP - 35

EP - 50

JO - Information Sciences

JF - Information Sciences

M1 - INS16166

ER -

An autoencoder wavelet based deep neural network with attention mechanism for multi-step prediction of plant growth

Abstract

Bibliographical note

Keywords

Access to Document

Fingerprint

Cite this