摘要
测试了不同分布密度常数对于模型分类的影响,并与标准BP神经网络的分类结果进行对比.仿真结果表明,所建立的PNN模型的分类平均准确率为96.00%,标准BP网络模型的分类平均准确率为87.33%,PNN模型的分类准确率高、泛化能力强、模型建立过程简单且训练无须太多样本.验证了将PNN用于移动学习策略分类方案的可行性.
The influences of different distribution density constants,upon a proposed model hased on probabilisticneural network（PNN）were tested. Its classification results were compared with those generated by a classificationmodel based on standard BP neural network（BPNN）. The simulation experiment results show that the averageaccuracy of the PNN model is 96.00%,while the one of BPNN model is 87.33%. PNN model has higher accuracyand better generalization ability. Its establishment process is simpler and it requires fewer training samples. Thefeasibility of applying PNN to the classification mobile learning strategies is verified.
出处
《河南科学》
2015年第11期1941-1945,共5页
Henan Science
基金
辽宁省教育厅科学研究一般项目(W2015015)
辽宁省社会科学基金资助项目(L14CYY022)
辽宁省社会科学基金重点项目(L15AYY001)
关键词
概率神经网络
移动学习策略
分类
probabilistic neural network
mobile learning strategy
classification
