A Comparative Analysis of Machine Learning Algorithms for Image Classification: Evaluating Performance

Image classification plays a crucial role in various applications, and selecting the most effective machine learning algorithm is essential for achieving accurate results. In this study, we conducted a comparative analysis of several well-known supervised machine learning techniques, including logistic regression, support vector machine (SVM), k-nearest neighbours (kNN), naïve Bayes, decision trees, random forest, AdaBoost, and artificial neural networks (ANN). To assess the performance of these algorithms, we utilised different fonts of the English alphabet as our dataset and performed the analysis using the R programming language. We evaluated the algorithms based on standard performance criteria, such as the area under the Receiver Operating Characteristic curve (ROC), accuracy, F1 score, precision, and recall. Our research findings demonstrated that the classification performance varied depending on the training size of the dataset. Notably, as the training size increased, neural networks exhibited superior performance compared to other machine learning techniques. Consequently, we conclude that neural networks and SVM are the most effective algorithms for image classification based on our study. By conducting this comprehensive analysis, we contribute valuable insights into selecting appropriate machine learning algorithms for image classification tasks. Our findings emphasise the significance of considering the training dataset size and highlight the advantages of neural networks and SVM in achieving high classification accuracy. This study provides valuable guidance for practitioners and researchers in choosing the most suitable machine learning algorithm for image classification, considering their specific requirements and dataset characteristics.