A modular LVQ neural network with fuzzy response integration for arrhythmia classification

Amezcua, Jonathan; Melín, Patricia

doi:10.1109/norbert.2014.6893884

Cited by 8 publications

(2 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In Section 5 the conclusions of this work are presented II. LVQ NETWROKS AND FUZZY SYSTEMS As mentioned before, LVQ is an adaptive classification method [1,3,8], its architecture is very similar to a competitive learning method, except that each output is associated with a class [12], Fig. 1 shows an example where the input dimension is 2, and the input space is divided into five clusters.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A new classification method based on LVQ neural networks and Fuzzy Logic

Amezcua

Melín

Castillo

2015

2015 Annual Conference of the North American Fuzzy Information Processing Society (NAFIPS) Held Jointly With 2015 5th World Con

Self Cite

View full text Add to dashboard Cite

In this paper, a new classification method based on LVQ neural networks and Fuzzy Logic is presented. This new fuzzy LVQ method (FuzzLVQ) mainly focuses on distances between the input vector and the cluster centers, randomly generated, thus the fuzzy system in the FuzzLVQ method is used to determine the shortest distance, and with this information, the cluster center can be approached to input vector if the classification was correct, or moved away in case of misclassification. This new method was tested for arrhythmia classification; the MIT-BIH arrhythmia dataset was used for this research, which consists of 15 classes. Experiments were conducted in a modular FuzzLVQ architecture with 5 modules, having 3 different classes of the dataset in each module.

show abstract

Section: Introductionmentioning

confidence: 99%

“…above shows an example of this process where input vectors p and a weight matrix w are defined; in the illustration an input vector p 1 is randomly taken, the distances between input vector p 1 and each of the cluster centers (weight matrix) w are obtained using(1), and the result is a vector with the distances.…”

mentioning

confidence: 99%