TY  - JOUR
AU  - Ibrahim, Hussain K. 
AU  - Guirguis, Shawkat K. 
AU  - Elsayed, Hend A. 
PY  - 2025
TI  - Nonnegative Matrix Factorization Features Extraction and Ensemble Methods Classifier for ECG Image Classification
JF  - Journal of Computer Science
VL  - 21
IS  - 4
DO  - 10.3844/jcssp.2025.918.927
UR  - https://thescipub.com/abstract/jcssp.2025.918.927
AB  - This study is a novel proposed classification system for ECG image classification. This proposed system uses nonnegative matrix factorization for feature extraction and ensemble methods for classification, and the results are compared with the features extracted by the principal component analysis, kernel principal component analysis, and independent component analysis. Although algorithms find image classification challenging, humans typically find it easy. Data processing might raise new issues during the decision-making process due to the huge and quick development of computers and information technology. In the subject of image classification, the researchers have encountered various challenges, particularly in identifying the image's best features that can provide a high degree of classification accuracy. The ultimate goal of this research is to optimize the image categorization process. There are six types of ensemble methods classifiers used in the proposed classification system. The six classification methods are the Adaboost M1 "Adaptive Boosting", Bagging Mea-Estimator, Logitboost, Gentleboost, Robustboost and Subspace. The results made for different numbers of feature extraction and As demonstrated by the experimental results, the bagging mea-estimator outperforms other ensemble methods for classification and the performance of the classifiers using features extracted from the nonnegative matrix factorization outperforms the performance using other feature extraction methods such as principal component analysis, kernel principal component analysis and independent component analysis. The performance of the classifiers using these methods is dependent on the number of features used.