Optimizing Support Vector Machine parameters using continuous Ant Colony Optimization

Support Vector Machines are considered to be excellent patterns classification techniques. The process of classifying a pattern with high classification accuracy counts mainly on tuning Support Vector Machine parameters which are the generalization error parameter and the kernel function parameter. Tuning these parameters is a complex process and may be done experimentally through time consuming human experience. To overcome this difficulty, an approach such as Ant Colony Optimization can tune Support Vector Machine parameters. Ant Colony Optimization originally deals with discrete optimization problems. Hence, in applying Ant Colony Optimization for optimizing Support Vector Machine parameters, which are continuous parameters, there is a need to discretize the continuous value into a discrete value. This discretization process results in loss of some information and, hence, affects the classification accuracy and seek time. This study proposes an algorithm to optimize Support Vector Machine parameters using continuous Ant Colony Optimization without the need to discretize continuous values for Support Vector Machine parameters. Seven datasets from UCI were used to evaluate the performance of the proposed hybrid algorithm. The proposed algorithm demonstrates the credibility in terms of classification accuracy when compared to grid search techniques. Experimental results of the proposed algorithm also show promising performance in terms of computational speed.