TY - JOUR
T1 - Cloud evolutionary computation system for advanced engineering analytics
AU - Chou, Jui Sheng
AU - Kosasih, Jeffisa Delaosia
AU - Chong, Wai K.
N1 - Funding Information:
Funding was provided by Ministry of Science and Technology, Taiwan (107-2221-E-011-035-MY3).
Publisher Copyright:
© 2021, The Author(s), under exclusive licence to Springer-Verlag London Ltd. part of Springer Nature.
PY - 2021
Y1 - 2021
N2 - The range of applications of artificial intelligence (AI) that is based on nature-inspired metaheuristics is rapidly increasing across various scientific fields as it is used to solve complex engineering problems. This work develops a cloud evolutionary machine learning system, called the nature-inspired metaheuristic optimization and prediction system (NiMOPS) that is composed of metaheuristic AI and web modules. The objective of the proposed system is to provide a user-friendly web analytics for making efficient, effective, and accurate predictions as solutions to engineering problems. For the purposes of web development, this work connects two programming languages, which are MATLAB and Java. A MATLAB Compiler is used to package the system into Java Archive (JAR) files, which provide the core modules for the development of the NiMOPS website using an integrated development environment (IDE). IDE compiles the JAR files, and web utilities (JavaScript, CSS, Servlet, and other utility files) to form the response-request connection between the user and the server. Therefore, the web-based system does not require the installation of an application by the users because they can access the cloud computing system ubiquitously with a browser or mobile device. Furthermore, it has many functions, including export—import file, train model, optimize prediction, save model and visualize results. Several case studies of this system, involving classification and regression problems, were examined. The analytic results of using the system to solve classification problems revealed that the system had a fault diagnosis accuracy of 96.5% and an accidental small fire accuracy of 52.4%. In solving regression problems, the root mean square errors were 28.58–68.82% better than those of previous methods. In particular, the proposed system performed multiple performance measures that were utilized in a regression analysis and were found to be more reliable evaluation metrics than used in elsewhere. The numerical experiments verified that cloud computing provides an innovative way to enable decision-makers to solve engineering problems.
AB - The range of applications of artificial intelligence (AI) that is based on nature-inspired metaheuristics is rapidly increasing across various scientific fields as it is used to solve complex engineering problems. This work develops a cloud evolutionary machine learning system, called the nature-inspired metaheuristic optimization and prediction system (NiMOPS) that is composed of metaheuristic AI and web modules. The objective of the proposed system is to provide a user-friendly web analytics for making efficient, effective, and accurate predictions as solutions to engineering problems. For the purposes of web development, this work connects two programming languages, which are MATLAB and Java. A MATLAB Compiler is used to package the system into Java Archive (JAR) files, which provide the core modules for the development of the NiMOPS website using an integrated development environment (IDE). IDE compiles the JAR files, and web utilities (JavaScript, CSS, Servlet, and other utility files) to form the response-request connection between the user and the server. Therefore, the web-based system does not require the installation of an application by the users because they can access the cloud computing system ubiquitously with a browser or mobile device. Furthermore, it has many functions, including export—import file, train model, optimize prediction, save model and visualize results. Several case studies of this system, involving classification and regression problems, were examined. The analytic results of using the system to solve classification problems revealed that the system had a fault diagnosis accuracy of 96.5% and an accidental small fire accuracy of 52.4%. In solving regression problems, the root mean square errors were 28.58–68.82% better than those of previous methods. In particular, the proposed system performed multiple performance measures that were utilized in a regression analysis and were found to be more reliable evaluation metrics than used in elsewhere. The numerical experiments verified that cloud computing provides an innovative way to enable decision-makers to solve engineering problems.
KW - Classification and regression for engineering analytics
KW - Cloud soft computing
KW - Evolutionary computing
KW - Human–machine interface
KW - Nature-inspired metaheuristic algorithm
KW - Optimized machine learning system
UR - http://www.scopus.com/inward/record.url?scp=85102088273&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85102088273&partnerID=8YFLogxK
U2 - 10.1007/s00366-020-01249-8
DO - 10.1007/s00366-020-01249-8
M3 - Article
AN - SCOPUS:85102088273
SN - 0177-0667
JO - Engineering with Computers
JF - Engineering with Computers
ER -