A Survey on Machine Learning Classification Techniques

ABSTRACT In the biomedical engineering, the assessment of the biological variations happening into the body of human is a challenge. Specifically, identifying the abnormal behavior of the human eye is really challenging due to the several complications in the process. The important part of the human eye is Retina, which can replicate the abnormal variations in the eye. Due to the requirement for disease identification techniques, the analysis of retinal image has gained sufficient significance in the research field. Since the diseases affect the human eye gradually, the identification of abnormal behavior using these techniques is really complex. These techniques are mostly reliant on manual intervention. But the success rate is quite low, since human observation is prone to error. These techniques must be highly accurate, since the treatment procedure varies for various abnormalities. Less accuracy may lead to fatal results due to wrong treatment. Hence, there should be such a automation technique, which give us high accuracy for disease identification applications of retina. This survey shows the study of different classification methods existed and their limitations which includes K-NN, SVM, Decision Trees, Naïve Bayes classifiers etc. But when we combine them to make an ensemble then classification accuracy can be improved.