A comparative study of general fuzzy min-max neural networks for pattern classification problems

Khuat, Thanh Tung; Gabrys, Bogdan

doi:10.1016/j.neucom.2019.12.090

Cited by 23 publications

(9 citation statements)

References 26 publications

(36 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, we do not tune the maximum hyperbox size for different datasets but use the same maximum hyperbox size parameter for all experimental datasets. If we use a small value of the maximum hyperbox size, then the GFMM model is expected to have a high classification performance, but the complexity of the model is also high [24]. In contrast, with large values of the maximum hyperbox size, the complexity of the GFMM model is low, but the classification performance is often not high.…”

Section: Datasets Parameter Settings and Performance Metricsmentioning

confidence: 99%

“…For several datasets, the encoded values in several dimensions contain only two values 0 and 1, so θ = 1 will ensure that the hyperboxes can be expanded to cover both of these extreme values. For the agglomerative learning algorithm, we used the "longest distance" [24] as a similarity measure and set σ = 0 so that the performance of the learning algorithm depends only on the values of θ. The sensitivity parameter γ in the membership function impacts the decreasing speed of the membership degrees for the numerical features.…”

Section: Datasets Parameter Settings and Performance Metricsmentioning

confidence: 99%

See 1 more Smart Citation

An in-depth comparison of methods handling mixed-attribute data for general fuzzy min-max neural network

Khuat¹,

Gabrys²

2020

Preprint

Self Cite

View full text Add to dashboard Cite

A general fuzzy min-max (GFMM) neural network is one of the efficient neuro-fuzzy systems for classification problems. However, a disadvantage of most of the current learning algorithms for GFMM is that they can handle effectively numerical valued features only. Therefore, this paper provides some potential approaches to adapting GFMM learning algorithms for classification problems with mixed-type or only categorical features as they are very common in practical applications and often carry very useful information. We will compare and assess three main methods of handling datasets with mixed features, including the use of encoding methods, the combination of the GFMM model with other classifiers, and employing the specific learning algorithms for both types of features. The experimental results showed that the target and James-Stein are appropriate categorical encoding methods for learning algorithms of GFMM models, while the combination of GFMM neural networks and decision trees is a flexible way to enhance the classification performance of GFMM models on datasets with the mixed features. The learning algorithms with the mixed-type feature abilities are potential approaches to deal with mixed-attribute data in a natural way, but they need further improvement to achieve a better classification accuracy. Based on the analysis, we also identify the strong and weak points of different methods and propose potential research directions.

show abstract

Section: Datasets Parameter Settings and Performance Metricsmentioning

confidence: 99%

Section: Datasets Parameter Settings and Performance Metricsmentioning

confidence: 99%

An in-depth comparison of methods handling mixed-attribute data for general fuzzy min-max neural network

Khuat¹,

Gabrys²

2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Fuzzy min-max neural networks are used to make the task easier and these implements fuzzy sets to accomplish tasks. 27,28…”

Section: Introductionmentioning

confidence: 99%

An evolutionary model for sleep quality analytics using fuzzy system

Hangaragi,

Nizampatnam,

Kaliyaperumal

et al. 2023

Proc Inst Mech Eng H

View full text Add to dashboard Cite

Electroencephalography (EEG) is a neuro signal reflecting brain activity. These signals provide information about brain activity, eye movements, and muscle tone, which can be used to determine the sleep stage. Categorizing sleep stages can be done manually by visually. Alternatively, automated algorithms can be developed using machine learning techniques to classify sleep stages based on signal features and patterns. This paper aims to automatically classify sleep stages based on extracted patterns from EEG signals. A fuzzy min-max neural network is proposed and implemented for sleep stage classification and clustering. The paper concludes that the fuzzy min-max neural network outperforms other tested methods in sleep stage classification. The models implemented in the study include K-Nearest Neighbor (KNN), Random Forest, Decision Tree, XGBoost, AdaBoost, Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), Convolutional Neural Network (CNN), and the fuzzy min-max classifier. The results indicate that the fuzzy classifier achieves the highest accuracy of 86%, followed by the CNN model with 81%. Among the machine learning algorithms, Random Forest with an accuracy of 55.46%, followed by XGBoost with 53.18%, surpassing the other algorithms used in the experiment. AdaBoost and Gaussian Naive Bayes both achieve an accuracy of 45.10%. Decision Tree, KNN, LDA, and QDA yield accuracies of 37.66%, 16.46%, 28.57%, and 29.5% respectively. These findings demonstrate the efficiency of the fuzzy min-max neural network and the superiority of the fuzzy classifier and CNN models in sleep stage classification, indicating their potential for accurate automated sleep stage analysis.

show abstract

“…Numerous scholars have integrated fuzzy theory and neural networks, and have designed neuro-fuzzy networks such as ANFIS, (18) IT2FNN, (19) and FMM. (20) Unlike traditional neural networks, neuro-fuzzy networks combine fuzzy logic (similar to human reasoning) and the learning ability of the neural network, which can automatically construct rules and reduce the number of learnable parameters in the network. However, parameter design in the network architecture is also a key factor related to the overall performance.…”

Section: Introductionmentioning

confidence: 99%

Arrhythmia Detection Using a Taguchi-based Convolutional Neuro-fuzzy Network

Li¹,

Jhang²,

Lin³

et al. 2022

Sensors and Materials

View full text Add to dashboard Cite

With improvements in the quality of life, people have paid increased attention to their health. According to the American Heart Association, cardiovascular disease was one of the leading causes of death globally as of 2016. Medical experts estimate that the worldwide annual number of people dying from cardiovascular disease will reach 23.6 million by 2030. Detecting heart arrhythmias effectively and quickly is critical for preventing cardiovascular disease. In this paper, a one-dimensional Taguchi-based convolutional neuro-fuzzy network (1D-TCNFN) for detecting arrhythmia in electrocardiograms (ECGs) is proposed. The proposed 1D-TCNFN adopts neuro-fuzzy instead of conventionally connected layers to reduce the number of learned parameters in the network. Four feature fusion methods, namely, global average pooling, global max pooling, channel average pooling, and channel max pooling, are employed in the 1D-TCNFN. For an increased detection accuracy, the Taguchi method was used to optimize the network architecture of the proposed 1D-TCNFN. In the experiments, the open Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) Arrhythmia Database was adopted to verify the performance of the proposed method for detecting 17 different arrhythmia signals.The proposed 1D-TCNFN exhibited a detection accuracy of 93.95% for the MIT-BIH Arrhythmia Database.

show abstract

A comparative study of general fuzzy min-max neural networks for pattern classification problems

Cited by 23 publications

References 26 publications

An in-depth comparison of methods handling mixed-attribute data for general fuzzy min-max neural network

An in-depth comparison of methods handling mixed-attribute data for general fuzzy min-max neural network

An evolutionary model for sleep quality analytics using fuzzy system

Arrhythmia Detection Using a Taguchi-based Convolutional Neuro-fuzzy Network

Contact Info

Product

Resources

About