基于遗传模拟退火算法的入侵检测特征选择研究

特征选择已经成为入侵检测的一个很重要的问题,它通过求解出有用的特征子集,去除冗余特征,来达到提高分类精度和效率的目的。提出了一种基于遗传模拟退火算法的特征选择算法,以基本遗传算法为基础,将遗传操作和模拟退火操作相结合,克服了一般遗传算法的不足,有效避免陷入局部最优并最终趋于全局优化,在入侵检测的标准数据集-KDD1999上的模拟实验证明：遗传模拟退火特征选择算法能够加快特征选择的速度;并且与当前典型的特征选择算法相比,具有更好的检测率。

Research on Feature Selection of Intrusion Detection based on Genetic and Simulated Annealing Algorithm

Feature selection is one of the most important problems in intrusion detection. It can work out the useful feature subset, eliminate redundant characteristic, and then, achieves the goals of improving classification precision and the efficiency. A feature selection arithmetic based on genetic and simulated annealing algorithm was put forward. On the basis of genetic algorithm,combines genetic operations and simulated annealing operations. Therefore,the algorithm which does well in avoiding some deficiencies...