Parameter Optimization of Concentration-Based Feature Construction Approaches
Ying Tan · 2015
This chapter presents a framework that can optimize the parameters of anti-spam models with heuristic swarm intelligence (SI) optimization algorithms and integrate various classifiers and feature extraction methods [225]. The related background information for proposing the framework is introduced first, as well as the knowledge of local concentration (LC)-based feature extraction approach and fireworks algorithm (FWA), which are selected as the representatives of anti-spam methods and SI optimization algorithms, respectively. The parameter optimization framework and its implementation on the selected methods are then described in detail. Finally, the experimental validation is given.