摘要
提高多聚脯氨酸二型结构的预测精度,可从两方面着手,一是结合遗传算法,形成遗传神经网络,努力使迭代往全局最优的方向进行;二是通过神经网络输入层,添加反映残基和预测中心位置距离的单元.结果表明,使用混合算法建立的新模型比仅用神经网络模型在预测精度上有明显提高,可从64.5%提高到70.1%.
A method is presented for prediction of polyproline type II based on a genetic algorithm neural network. High predicting accuracy is improved at two aspects:one is from the algorithm combined with genetic algorithm; the another is from the input layer.The result shows that the method improves the predicting accuracy,from BP’s 64.5% to Genetic Neural Network’s 70.1% .
出处
《江南大学学报（自然科学版）》
CAS
2005年第3期244-247,251,共5页
Joural of Jiangnan University (Natural Science Edition)
