Application of SVM, FFNNs, k-NN and Their Ensembles for Identifying Functionally Reliable Systems
Active informatization of various spheres of human activity requires increasingly widespread use of information systems. Along with the growing need for their application, the demands on the systems themselves are also rising. Some of these demands can be addressed through technical improvements; ho...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-03-01
|
| Series: | Axioms |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-1680/14/4/237 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Active informatization of various spheres of human activity requires increasingly widespread use of information systems. Along with the growing need for their application, the demands on the systems themselves are also rising. Some of these demands can be addressed through technical improvements; however, there are aspects for which this alone may not suffice. One such requirement is functional stability. While it is technically possible to ensure functional stability, a number of indicators and criteria have been developed for assessing it. However, applying these indicators in real-world conditions requires significant computational resources. Therefore, there is a need to develop more optimized methods to evaluate whether a system is functionally stable or to improve existing ones. Recently, interest in machine learning methods as a means of optimizing various computations has grown significantly. Accordingly, the question arises as to whether machine learning can be applied to assess the functional stability of information systems. In this study, we investigate the application of some popular classification methods—SVM, FFNNs, k-NN and their ensembles—to determine compliance with one of the requirements for the structure of information systems, which helps evaluate whether the system is functionally stable. |
|---|---|
| ISSN: | 2075-1680 |