S. Pourmohammad Azizi, A. Neisy, Sajad Ahmad Waloo
{"title":"A Dynamical Systems Approach to Machine Learning","authors":"S. Pourmohammad Azizi, A. Neisy, Sajad Ahmad Waloo","doi":"10.1142/s021987622350007x","DOIUrl":null,"url":null,"abstract":"Employing various mathematical tools in machine learning is crucial since it may enhance the learning problem’s efficiency. Dynamic systems are among the most effective tools. In this study, an effort is made to examine a kind of machine learning from the perspective of a dynamic system, i.e., we apply it to learning problems whose input data is a time series. Using the discretization approach and radial basis functions, a new data set is created to adapt the data to a dynamic system framework. A discrete dynamic system is modeled as a matrix that, when multiplied by the data of each time, yields the data of the next time, or, in other words, can be used to predict the future value based on the present data, and the gradient descent technique was used to train this matrix. Eventually, using Python software, the efficacy of this approach relative to other machine learning techniques, such as neural networks, was analyzed.","PeriodicalId":54968,"journal":{"name":"International Journal of Computational Methods","volume":null,"pages":null},"PeriodicalIF":1.4000,"publicationDate":"2023-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"International Journal of Computational Methods","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.1142/s021987622350007x","RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENGINEERING, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0
Abstract
Employing various mathematical tools in machine learning is crucial since it may enhance the learning problem’s efficiency. Dynamic systems are among the most effective tools. In this study, an effort is made to examine a kind of machine learning from the perspective of a dynamic system, i.e., we apply it to learning problems whose input data is a time series. Using the discretization approach and radial basis functions, a new data set is created to adapt the data to a dynamic system framework. A discrete dynamic system is modeled as a matrix that, when multiplied by the data of each time, yields the data of the next time, or, in other words, can be used to predict the future value based on the present data, and the gradient descent technique was used to train this matrix. Eventually, using Python software, the efficacy of this approach relative to other machine learning techniques, such as neural networks, was analyzed.
期刊介绍:
The purpose of this journal is to provide a unique forum for the fast publication and rapid dissemination of original research results and innovative ideas on the state-of-the-art on computational methods. The methods should be innovative and of high scholarly, academic and practical value.
The journal is devoted to all aspects of modern computational methods including
mathematical formulations and theoretical investigations;
interpolations and approximation techniques;
error analysis techniques and algorithms;
fast algorithms and real-time computation;
multi-scale bridging algorithms;
adaptive analysis techniques and algorithms;
implementation, coding and parallelization issues;
novel and practical applications.
The articles can involve theory, algorithm, programming, coding, numerical simulation and/or novel application of computational techniques to problems in engineering, science, and other disciplines related to computations. Examples of fields covered by the journal are:
Computational mechanics for solids and structures,
Computational fluid dynamics,
Computational heat transfer,
Computational inverse problem,
Computational mathematics,
Computational meso/micro/nano mechanics,
Computational biology,
Computational penetration mechanics,
Meshfree methods,
Particle methods,
Molecular and Quantum methods,
Advanced Finite element methods,
Advanced Finite difference methods,
Advanced Finite volume methods,
High-performance computing techniques.