摘要：
传统BP神经网络存在容易陷入局部最优、收敛速度慢等缺点。为此,根据人工蜂群算法全局寻优以及群体智能的特点,在初始化神经网络参数时,将神经网络的误差作为人工蜂群算法的适应度,选择适应度最好的一组参数作为神经网络的权值和阈值,避免神经网络陷入局部最优和收敛速度慢的问题。将人工蜂群优化的BP神经网络模型应用于入侵检测中,仿真实验结果表明,优化后的网络模型可加快收敛速度,提高检测精度。
关键词:
人工蜂群算法,
BP神经网络,
入侵检测,
遗传算法,
全局寻优
Abstract: The existence of traditional BP neural network is easy to fall into local optimum,slow convergence and other shortcomings.According to the features of global optimization and swarm intelligence of Artificial Bee Colony(ABC) algorithm,in the neural network parameter initialization,this paper uses the deviation of the neural network as a fitness of ABC algorithm,selects the best fitness of a set of parameters as a nerve power networks and thresholds.Doing so can avoid falling into local optimum neural network and slow convergence problem.The BP neural network model of ABC optimization applied to intrusion detection.Simulation results show that the network model is optimized to accelerate the convergence rate and improve the detection accuracy.
Key words:
Artificial Bee Colony(ABC) algorithm,
BP neural network,
intrusion detection,
genetic algorithm,
global optimization
中图分类号: