【摘要】 针对传统特征选择算法在大数据集下选择能力差、选择效率低等问题，提出结合深度神经网络的特征选择算法实现高维数据的数据降维。首先，将特征选择嵌入到深度神经网络中去，利用L_2,1范数对输入层与第一隐层之间的参数进行约束；其次，在训练模型过程中，利用梯度下降算法对模型的参数进行更新，训练结束后，利用这部分参数的稀疏性来选择特征；最后，在人造验证数据集、公共数据集和实际应用的工业质检数据集上对比了目前性能最佳的4种特征选择算法，实验结果证明了方法的有效性和优越性。更多还原

【Abstract】 In order to highlight the data dimensionality reduction ability of feature selection, this paper proposes a feature selection algorithm combined with deep neural network to achieve data dimensionality reduction in view of the poor selection ability and low selection efficiency of traditional feature selection algorithms in large datasets. Firstly, feature selection is embedded into the deep neural network, and the parse norm is used to constrain the parameters between the input layer and the first hidden layer. Secondly, in the process of training the model, it uses the gradient descent algorithm to update the parameters of the model. After training, the sparsity of this part of the parameters is used to select features. Finally, four feature selection algorithms with the best performance are compared on artificial validation data sets, public data sets, and practical industrial quality inspection datasets. The experimental results prove the effectiveness and superiority of the method.更多还原