【摘要】 提出了一种改进的ReliefF-BPNN分类模型。该模型使用ReliefF算法和交互增益权重，来最大程度地保留相关特征与交互特征；同时在BP神经网络模型的误差函数中加入正则化项防止过拟合。实验表明，改进的ReliefF-BPNN在大多数数据集上的分类准确率高于90%，其精度相对于其他传统模型更高。更多还原

【Abstract】 In this paper, an improved ReliefF-BPNN classification model is proposed. It uses the ReliefF algorithm and interaction gain weights to maximize the retention of correlation and interaction features. Meanwhile, a regularization term is added to the error function of BP neural network model to prevent overfitting. Experiments show that the classification accuracy of the improved ReliefF-BPNN is higher than 90% on most data sets, and its accuracy is higher than that of other traditional models.更多还原