A prediction model of rectums perceptive function reconstruction based on SVM optimized by ACO

作者：Zan Peng; Ai Yutao; Zhao Jie; Shao Yong^*

来源：Communications in Computer and Information Science, 2014, 461: 121-128.

• Artificial anal sphincters
• Classification accuracy
• Function reconstruction
• Nonlinear and non-stationary signals
• Optimal wavelets
• Prediction accuracy
• Prediction model
• RBF kernel function

In order to solve the problem that rectal pressure signal typically has complicate nonlinear and non-stationary signal characteristics. In this paper, our model uses wavelet packet to extract feature vectors and classifies them based on SVM. In our model, we choose RBF kernel function which has been proved has best discrimination for rectal pressure signals in the experiments. At the same time, in order to acquire higher accuracy, ACO is introduced to find the optimal value of parameters. We compare the prediction accuracy of SVM with different kernel functions. The results show that when we input parameters which were optimized by ACO into SVM, it acquires the highest classification accuracy and has preferable predication performance. So we can conclude that the proposed method is an effective way to rebuild patients’ rectal perception function.

• 出版日期2014
• 单位上海大学