基于遗传BPN的舞弊甄别的系统设计

System Design on Discrimination of Fraud Based on Genetic BPN

近几年,计算机领域的发展使财务报告舞弊更加复杂、隐蔽性更好,再加上传统统计技术的落后,使得激增的数据背后隐藏的信息正在被人们所忽略。通过将遗传BPN技术应用于舞弊性财务报告的识别,以及对舞弊甄别系统的分析和设计,使舞弊甄别的主要功能模块通过计算机来实现,得出由于遗传算法能够收敛到全局最优解,而且遗传算法的鲁棒性强,将遗传算法与神经网络结合起来是很有意义的,不仅能发挥神经网络的泛化的映射能力,而且使神经网络具有很快的收敛性以及较强的学习能力,从而提高了舞弊甄别的效率,缩短了神经网络训练的时间等结论,进一步证明了遗传BP神经网络技术在会计舞弊甄别中的科学性、准确性、可行性。

In recent years, the development in computer field makes fraud of financial report more complicated and hidden, in addition, backward traditional statistic technique makes the hidden information behind the explosive data ignored by people. Through applying genetic BPN technique to distinguish fraudulent financial statements, and analyzing and designing discrimination system of fraud, we will realize the main functional mode of fraud diseriminafon, and conclude that the combination between genetic algorithm and neural network is siguifieant because genetic algorithm can collect global optimal solution and robustness of genetic algorithm is strong, which can play extensive mapping ability of neural network and put neural network with rapid convergence and strong learning ability, correspondingly, increase the efficiency of fraud discrimination, shorten the training time of neural network, and further prove scientific, accurate and practical characteristics of genetic BP neural network technique in accounting fraud discrimination.