Learning vector quantization neural networks were trained to classify healthy and arrhythmia cases from the UCI ECG dataset, though specific quantitative performance metrics were not reported.
The aim of this study is to apply learning vector quantization (LVQ) neural networks to classify arrhythmia from the Electrocardiogram (ECG) dataset. LVQ classification algorithms do not approximate density functions of class samples but directly define class boundaries based on prototypes, a nearest-neighbor rule and a winner-takes-it-all paradigm. It has a superior performance over back-propagation (BP) method in the sense of minimizing the classification errors while maintaining rapid convergence. First, the principal component analysis is used to reduce dimensionality of the input features and increase the distinguishing capability. Then, six LVQ neural networks are trained to classify each case into ¿healthy¿ and ¿arrhythmia¿ classes. The networks are trained and tested for the UCI ECG arrhythmia dataset. This dataset is a good environment to test classifiers as it is incomplete and ambiguous bio-signal data from multiple patients. The classification performance of each algorithm is evaluated using four measures; sensitivity, specificity, classification accuracy and time taken to build the system. Experimental results recommend using LVQ algorithm for a more extended research regarding this topic.
Alaa M. Elsayad (Tue,) conducted a other in Arrhythmia. Learning vector quantization (LVQ) neural networks vs. Back-propagation (BP) method was evaluated on Classification performance (sensitivity, specificity, classification accuracy, and time taken). Learning vector quantization neural networks were trained to classify healthy and arrhythmia cases from the UCI ECG dataset, though specific quantitative performance metrics were not reported.