Mathematical Composite Models, a Path to Solve Research Complex Problems

Advanced Topics in Optoelectronics, Microelectronics, and Nanotechnologies XI

Menabil²

2023

Searching the intelligent aspects and identifying the rules in the huge amount of data automatically created nowadays is an appropriate method to explore and understand a world that is becoming more complex. Fortunately, the means of investigations are synchronous with the actual level of technological development. One might remark that, computer science being a rapid development field, it became an inspiring environment for other domains of science. The paper presents data mining techniques, original implementations of the linear/quadratic regression and of the K-means clustering methods and two machine learning models using the “Microsoft Machine Learning Studio (classic)” software. Some of the conclusions are: the data input process is simple and flexible, making it possible to access a wide range of data sources; there is an important library of algorithms that can be easily included in the experiment; data control may be thoroughly done using various methods, including visualizations; the machine learning application offers reliable results; the output consists of several types of reusable ‘objects’: trained models, data in various formats, web services. The paper is an important stage on the path of a trust development process which is intended to finish by considering machine learning instruments as common instruments in engineering.

“…Inter-Domain Approaches with Applications in Engineering and Economy", under the supervision of the National Authority for Scientific Research, Romania, 2010-2012 13 .…”

Section: Discussionmentioning

confidence: 99%

Machine learning tests for data mining in engineering

Advanced Topics in Optoelectronics, Microelectronics, and Nanotechnologies XI

Menabil²

2023

“…The ideas presented in this scientific paper are inspired by the results of the ID1223 -"Computer Aided Advanced Studies in Applied Elasticity from an Interdisciplinary Perspective" Scientific Research Project, under the supervision of the National University Research Council (CNCSIS), Romania, 2007-2010 7 and by the follow-up scientific research project "Mathematical Models for Inter-Domain Approaches with Applications in Engineering and Economy", under the supervision of the National Authority for Scientific Research, Romania, 2010-2012 8 .…”

Section: Acknowledgementsmentioning

confidence: 99%

Class template: primary approach for the flexible dynamic memory allocation in hybrid modelling software development

Advanced Topics in Optoelectronics, Microelectronics, and Nanotechnologies XI

Pescaru²

2023

Hybrid modeling is an intelligent knowledge acquiring strategy that becomes very important in the nowadays complex conditions and severe constraints. Hybrid modeling requires several types of 'building blocks' or basic components: standalone efficient solvers, interfaces between the data sources, mathematical libraries etc. When a new model must be conceived, new components must be swiftly developed. The short access time to large amount of data is a basic requirement which was a constant concern of ours along the time. Our latest solution is based on class templates used to dynamically allocate the memory by creating circular doubly linked lists and other upper level data structures. Facilities of the C++ Standard Template Library were also studied. We may consider the template classes as ‘stem cells’ in the software development process. Several cycles consisting of thorough analysis plus development of software demonstrators meant to clarify practical aspects led to a firm foundation and a strategy to conceive and rapidly develop effective software components for hybrid models, regardless the field of science.

“…The scientific paper is a quest for intelligent software solutions starting from the ID1223 -"Computer Aided Advanced Studies in Applied Elasticity from an Interdisciplinary Perspective" Scientific Research Project, under the supervision of the National University Research Council (CNCSIS), Romania, 2007-2010 11 , which was followed by the scientific research project "Mathematical Models for Inter-Domain Approaches with Applications in Engineering and Economy", under the supervision of the National Authority for Scientific Research, Romania, 2010-2012 12 .…”

Section: Acknowledgementsmentioning

confidence: 99%

Data management in original software development for hybrid models

Advanced Topics in Optoelectronics, Microelectronics, and Nanotechnologies XI

Menabil²

2023

Huge amount of data is generated nowadays by many and various sources and their management is a problem of high importance in science, especially in research and development. Complex phenomena can be accurately modeled using various sources of large amount of data which either cross check the models, or improve their accuracy. Data management is an important issue in actual modeling context, which must be solved at both, concept level and applications' layer. The paper presents a ‘checklist’ of questions to be answered when the data management layer of a new computer aided model must be conceived and implemented. Moreover, finding the rules of the ‘game’, i.e. the best policies for the rapid development of the data management foundation of the models leads to a valuable knowhow in the rapid composite models’ development or even in hybrid models’ development. The paper presents the conclusions of a thorough study of the MySQL, Oracle, Microsoft SQL Server and MariaDB RDBMSs. A subsequent study will approach MongoDB, Amazon DynamoDB and Databricks in order to synchronize the models’ development requirements with the actual cloud approaches: IaaS, PaaS and SaaS. To conclude, in the huge volume of data actual conditions, an intelligent data persistency strategy within a model is a goal for a smart and visionary analyst, who is, according to Blaise Pascal, a “thinking reed”, i.e. “Man is but a reed, the most feeble thing in nature, but he is a thinking reed”.