Analyzing various Machine Learning Algorithms with SMOTE and ADASYN for Image Classification having Imbalanced Data
Yakshit, Gagandeep Kaur, Veerpal Kaur, Yashika Sharma, Vishnu Bansal
Abstract
Oversampling is a strategy employed in machine learning to handle imbalanced datasets by creating copies of the minority class instances to balance the dataset, thus reducing bias and enhancing the accuracy of the model. The work presents various Machine Learning techniques to classify images from an imbalanced dataset. In this paper various oversampling techniques such as ADASYN and SMOTE are blended with the classification algorithms i.e., SVM and CNN with SVM in order to balance imbalanced datasets. The experimentation is performed on the Google Colab using different Machine Learning techniques: SVM, and (CNN and SVM). The results of the experimentation suggests that the amalgamation of SVM and CNN is better than the SVM and SMOTE is better than ADASYN on the basis of performance matrices such as recall, precision, F1 score.