并行神经元集成在风险评估中的应用

Parallel algorithms based on neural network ensemble for risk assessment

研究了信息安全风险评估的并行求解方法.以神经元集成为出发点,借助于多个神经元网络并行逼近风险评估值,以降低评估时间和提高评估精度.提出了并行神经元集成风险评估算法,并应用到实例中,实验表明,并行算法能够有效地减少评估花费的时间和增加风险评估准确性.

The article studies the parallelizing method for information security risk assessment. Based on neural network ensemble, the paper establishes the ways of parallel training by using some diversity neural network, and gives a numerical simulation example to illustrate these methods. The experimental results show that parallel algorithm can decrease the computing time and yield higher accuracy.