Analysis on Misclassification in Existing Contraction of Fuzzy Min–max Models
Fuzzy min–max (FMM) neural network is one of the most powerful models for pattern classification. Various models have been introduced based on FMM model to improve the classification performance. However, the misclassification of the contraction process is a crucial issue that has to be handled in F...
| Main Authors: | , , , , |
|---|---|
| Format: | Conference or Workshop Item |
| Language: | English |
| Published: |
Springer International Publishing
2020
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/25535/ http://umpir.ump.edu.my/id/eprint/25535/ http://umpir.ump.edu.my/id/eprint/25535/1/Analysis%20on%20Misclassification%20in%20Existing.pdf |
| Summary: | Fuzzy min–max (FMM) neural network is one of the most powerful models for pattern classification. Various models have been introduced based on FMM model to improve the classification performance. However, the misclassification of the contraction process is a crucial issue that has to be handled in FMM models to improve classification accuracy. Hence, this research aims to analyse the existence and execution procedure of addressing the misclassification of the contraction in the current FMM models. In this manner, practitioners and researchers are aided in selecting the convenient model that can address the misclassification of the contraction and improve the performance of models in producing accurate classification results. A total of 15 existing FMM models are identified and analysed in terms of the contraction problem. Results reveal that only five models can address the contraction misclassification problem. However, these models suffer from serious limitations, including the inability to detect all overlap cases, and increasing the network structure complexity. A new model is thus needed to address the specified limitations for increasing the pattern classification accuracy. |
|---|