Improvement of LSTM-Based Forecasting with NARX Model through Use of an Evolutionary Algorithm
The reported work aims to improve the performance of LSTM-based (Long Short-Term Memory) forecasting algorithms in cases of NARX (Nonlinear Autoregressive with eXogenous input) models by using evolutionary search. The proposed approach, ES-LSTM, combines a two-membered ES local search procedure (2ME...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-09-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/11/18/2935 |
| _version_ | 1797489071719710720 |
|---|---|
| author | Cătălina Lucia Cocianu Cristian Răzvan Uscatu Mihai Avramescu |
| author_facet | Cătălina Lucia Cocianu Cristian Răzvan Uscatu Mihai Avramescu |
| author_sort | Cătălina Lucia Cocianu |
| collection | DOAJ |
| description | The reported work aims to improve the performance of LSTM-based (Long Short-Term Memory) forecasting algorithms in cases of NARX (Nonlinear Autoregressive with eXogenous input) models by using evolutionary search. The proposed approach, ES-LSTM, combines a two-membered ES local search procedure (2MES) with an ADAM optimizer to train more accurate LSTMs. The accuracy is measured from both error and trend prediction points of view. The method first computes the learnable parameters of an LSTM, using a subset of the training data, and applies a modified version of 2MES optimization to tune them. In the second stage, all available training data are used to update the LSTM’s weight parameters. The performance of the resulting algorithm is assessed versus the accuracy of a standard trained LSTM in the case of multiple financial time series. The tests are conducted on both training and test data, respectively. The experimental results show a significant improvement in the forecasting of the direction of change without damaging the error measurements. All quality measures are better than in the case of the standard algorithm, while error measures are insignificantly higher or, in some cases, even better. Together with theoretical consideration, this proves that the new method outperforms the standard one. |
| first_indexed | 2024-03-10T00:11:19Z |
| format | Article |
| id | doaj.art-98a2beae81ba4c4da600c904b6331b19 |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-10T00:11:19Z |
| publishDate | 2022-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-98a2beae81ba4c4da600c904b6331b192023-11-23T15:59:08ZengMDPI AGElectronics2079-92922022-09-011118293510.3390/electronics11182935Improvement of LSTM-Based Forecasting with NARX Model through Use of an Evolutionary AlgorithmCătălina Lucia Cocianu0Cristian Răzvan Uscatu1Mihai Avramescu2Department of Economic Informatics and Cybernetics, Bucharest University of Economic Studies, 10552 Bucharest, RomaniaDepartment of Economic Informatics and Cybernetics, Bucharest University of Economic Studies, 10552 Bucharest, RomaniaDepartment of Economic Informatics and Cybernetics, Bucharest University of Economic Studies, 10552 Bucharest, RomaniaThe reported work aims to improve the performance of LSTM-based (Long Short-Term Memory) forecasting algorithms in cases of NARX (Nonlinear Autoregressive with eXogenous input) models by using evolutionary search. The proposed approach, ES-LSTM, combines a two-membered ES local search procedure (2MES) with an ADAM optimizer to train more accurate LSTMs. The accuracy is measured from both error and trend prediction points of view. The method first computes the learnable parameters of an LSTM, using a subset of the training data, and applies a modified version of 2MES optimization to tune them. In the second stage, all available training data are used to update the LSTM’s weight parameters. The performance of the resulting algorithm is assessed versus the accuracy of a standard trained LSTM in the case of multiple financial time series. The tests are conducted on both training and test data, respectively. The experimental results show a significant improvement in the forecasting of the direction of change without damaging the error measurements. All quality measures are better than in the case of the standard algorithm, while error measures are insignificantly higher or, in some cases, even better. Together with theoretical consideration, this proves that the new method outperforms the standard one.https://www.mdpi.com/2079-9292/11/18/2935evolutionary strategiesLSTM neural networksFourier filteringtime seriesdata standardization |
| spellingShingle | Cătălina Lucia Cocianu Cristian Răzvan Uscatu Mihai Avramescu Improvement of LSTM-Based Forecasting with NARX Model through Use of an Evolutionary Algorithm Electronics evolutionary strategies LSTM neural networks Fourier filtering time series data standardization |
| title | Improvement of LSTM-Based Forecasting with NARX Model through Use of an Evolutionary Algorithm |
| title_full | Improvement of LSTM-Based Forecasting with NARX Model through Use of an Evolutionary Algorithm |
| title_fullStr | Improvement of LSTM-Based Forecasting with NARX Model through Use of an Evolutionary Algorithm |
| title_full_unstemmed | Improvement of LSTM-Based Forecasting with NARX Model through Use of an Evolutionary Algorithm |
| title_short | Improvement of LSTM-Based Forecasting with NARX Model through Use of an Evolutionary Algorithm |
| title_sort | improvement of lstm based forecasting with narx model through use of an evolutionary algorithm |
| topic | evolutionary strategies LSTM neural networks Fourier filtering time series data standardization |
| url | https://www.mdpi.com/2079-9292/11/18/2935 |
| work_keys_str_mv | AT catalinaluciacocianu improvementoflstmbasedforecastingwithnarxmodelthroughuseofanevolutionaryalgorithm AT cristianrazvanuscatu improvementoflstmbasedforecastingwithnarxmodelthroughuseofanevolutionaryalgorithm AT mihaiavramescu improvementoflstmbasedforecastingwithnarxmodelthroughuseofanevolutionaryalgorithm |