论文部分内容阅读
采用误差反传前向人工神经网络(artificial neural network,ANN)建立了21种2-(4-取代-苯基)-3-异噻唑啉酮类化合物的结构与其抗菌活性之间的定量关系模型(ANN模型),以21种3-异噻唑啉酮类化合物的量子化学参数和拓扑指数作为输入、抗菌活性作为输出,所构建网络模型的交叉检验相关系数为0.991 6、标准偏差为0.080 1、残差绝对值≤0.221,应用于外部预测集,预测集相关系数为0.973 1;而多元线性回归(multiple linearregression,MLR)法模型的相关系数为0.841 8、标准偏差为0.303 9、残差绝对值≤0.636。结果表明,ANN模型获得了比MLR模型更好的拟合效果。
The quantitative relationship between the structure of 21 2- (4-substituted-phenyl) -3-isothiazolinones and their antibacterial activity was established by artificial neural network (ANN) (ANN model). With the input of 21 kinds of 3-isothiazolinone compounds’ quantum chemical parameters and topological index, the antibacterial activity was taken as the output. The correlation coefficient of the cross-test of the constructed network model was 0.991 6, the standard deviation was 0.080 1, The residual absolute value was ≤0.221, which was applied to the external prediction set. The correlation coefficient of the prediction set was 0.973 1. The correlation coefficient of the multiple linear regression (MLR) method was 0.841 8 with the standard deviation of 0.303 9. The residual absolute value ≤0.636. The results show that ANN model has better fitting effect than MLR model.