Evaluation of Heuristics for Taken’s Theorem Hyper-Parameters Optimization in Time Series Forecasting Tasks
This study compares three methods for optimizing the hyper-parameters <i>m</i> (embedding dimension) and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>τ</mi></semantics></math>...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-07-01
|
| Series: | Engineering Proceedings |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2673-4591/39/1/71 |
| _version_ | 1797580302267187200 |
|---|---|
| author | Rodrigo Hernandez-Mazariegos Jose Ortiz-Bejar Jesus Ortiz-Bejar |
| author_facet | Rodrigo Hernandez-Mazariegos Jose Ortiz-Bejar Jesus Ortiz-Bejar |
| author_sort | Rodrigo Hernandez-Mazariegos |
| collection | DOAJ |
| description | This study compares three methods for optimizing the hyper-parameters <i>m</i> (embedding dimension) and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>τ</mi></semantics></math></inline-formula> (time delay) from Taken’s Theorem for time-series forecasting to train a Support Vector Regression system (SVR). Firstly, we use a method which utilizes Mutual Information for optimizing <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>τ</mi></semantics></math></inline-formula> and a technique referred to as “Dimension Congruence” to optimize <i>m</i>. Secondly, we employ a grid search and random search, combined with a cross-validation scheme, to optimize <i>m</i> and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>τ</mi></semantics></math></inline-formula> hyper-parameters. Lastly, various real-world time series are used to analyze the three proposed strategies. |
| first_indexed | 2024-03-10T22:48:12Z |
| format | Article |
| id | doaj.art-4ff031625bdd4fa3a6d6504ea583ce8c |
| institution | Directory Open Access Journal |
| issn | 2673-4591 |
| language | English |
| last_indexed | 2024-03-10T22:48:12Z |
| publishDate | 2023-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Engineering Proceedings |
| spelling | doaj.art-4ff031625bdd4fa3a6d6504ea583ce8c2023-11-19T10:31:11ZengMDPI AGEngineering Proceedings2673-45912023-07-013917110.3390/engproc2023039071Evaluation of Heuristics for Taken’s Theorem Hyper-Parameters Optimization in Time Series Forecasting TasksRodrigo Hernandez-Mazariegos0Jose Ortiz-Bejar1Jesus Ortiz-Bejar2Facultad de Ciencias Físico-Matemáticas “Mat. Luis Manuel Rivera Gutiérrez”, UMSNH, Avenida Universidad 100, Villa Universidad, 58060 Morelia, Michoacán, MexicoDivisión de Estudios de Posgrado de la Facultad de Ingeniería Eléctrica, UMSNH, Building “Ω2” Ciudad Universitaria, Francisco J. Múgica S/N, 58030 Morelia, Michoacán, MexicoFacultad de Ciencias Físico-Matemáticas “Mat. Luis Manuel Rivera Gutiérrez”, UMSNH, Avenida Universidad 100, Villa Universidad, 58060 Morelia, Michoacán, MexicoThis study compares three methods for optimizing the hyper-parameters <i>m</i> (embedding dimension) and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>τ</mi></semantics></math></inline-formula> (time delay) from Taken’s Theorem for time-series forecasting to train a Support Vector Regression system (SVR). Firstly, we use a method which utilizes Mutual Information for optimizing <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>τ</mi></semantics></math></inline-formula> and a technique referred to as “Dimension Congruence” to optimize <i>m</i>. Secondly, we employ a grid search and random search, combined with a cross-validation scheme, to optimize <i>m</i> and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>τ</mi></semantics></math></inline-formula> hyper-parameters. Lastly, various real-world time series are used to analyze the three proposed strategies.https://www.mdpi.com/2673-4591/39/1/71Taken’s Theoremtime-seriesSVR forecastingmutual informationdimension congruencerandom search |
| spellingShingle | Rodrigo Hernandez-Mazariegos Jose Ortiz-Bejar Jesus Ortiz-Bejar Evaluation of Heuristics for Taken’s Theorem Hyper-Parameters Optimization in Time Series Forecasting Tasks Engineering Proceedings Taken’s Theorem time-series SVR forecasting mutual information dimension congruence random search |
| title | Evaluation of Heuristics for Taken’s Theorem Hyper-Parameters Optimization in Time Series Forecasting Tasks |
| title_full | Evaluation of Heuristics for Taken’s Theorem Hyper-Parameters Optimization in Time Series Forecasting Tasks |
| title_fullStr | Evaluation of Heuristics for Taken’s Theorem Hyper-Parameters Optimization in Time Series Forecasting Tasks |
| title_full_unstemmed | Evaluation of Heuristics for Taken’s Theorem Hyper-Parameters Optimization in Time Series Forecasting Tasks |
| title_short | Evaluation of Heuristics for Taken’s Theorem Hyper-Parameters Optimization in Time Series Forecasting Tasks |
| title_sort | evaluation of heuristics for taken s theorem hyper parameters optimization in time series forecasting tasks |
| topic | Taken’s Theorem time-series SVR forecasting mutual information dimension congruence random search |
| url | https://www.mdpi.com/2673-4591/39/1/71 |
| work_keys_str_mv | AT rodrigohernandezmazariegos evaluationofheuristicsfortakenstheoremhyperparametersoptimizationintimeseriesforecastingtasks AT joseortizbejar evaluationofheuristicsfortakenstheoremhyperparametersoptimizationintimeseriesforecastingtasks AT jesusortizbejar evaluationofheuristicsfortakenstheoremhyperparametersoptimizationintimeseriesforecastingtasks |