The model overcomes some flaws of popular learning algorithm such as local minima, slow convergence and initialized values.

• 该模型克服了传统入侵检测系统所存在的局部极小、收敛速度慢、初值敏感性等问题。