Perturbation based efficient crow search optimized FLANN for system identification: A novel approach

Abstract

Designing an efficient classifier is a tough task as it must be suitable for solving maximum real life problems with high accuracy and less error rate. In this paper, a novel functional link neural network based system identification model is developed to solve the classification problem of data mining. To increase the accuracy of the model and for an optimized performance, an enhanced crow search algorithm (CSA) with perturbation has been introduced. This enhanced version of CSA based model avoids premature convergence and stagnation in classical CSA, by introducing the new neighbourhood searching operation through perturbation. Experimental results reveal that the proposed model outperforms several other standard models in terms of accuracy and error rate.