Feature Selection and Parameters Optimization of Support Vector Machines Based on Hybrid Glowworm Swarm Optimization for Classification of Diabetic Retinopathy

Diabetic Retinopathy (DR) has been a leading cause of blindness in case of human beings falling between the ages of 20 and 74 years. This will have a major influence on both the patient and the society as it can normally influence the humans in their gainful years. An early DR detection is quite challenging as it may not be detected by humans. There are several techniques and algorithms that have been established for detecting the DR. These techniques have been facing problems to achieve effective sensitivity, accuracy, and specificity. In order to overcome all these problems, the work has proposed one more such effective algorithm for image processing in order to increase the efficiency and also identify easily the DR diseases. A major challenge in the task is the automatic detection of the microaneurysms. In this work, the Support Vector Machine (SVM) parameters optimized with Glowworm Swarm Optimization (GSO) and Genetic Algorithm (GA) is used to classify the DR. Because the SVM parameter C and γ to control the performance of the classifier. For this work, the SVMs get fused with the hybrid GSO-GA along with the feature chromosomes that are generated that will thereby direct the GA search to a straight line of the error of optimal generalization in their super parameter space. This GSO algorithm will not have memory and the glow worms will not retain any information in memory. The results of the experiment prove that this method had achieved a better performance.