Application of Autoencoder in dimension Reduction of Dummy Response

To solve the problem that the dimensions of the dummy response data are too high to be applied directly, an autoencoder based method for dimension reduction of the dummy response data is proposed in this paper. Firstly, the data characteristics of the high dimensional data of the mechanical response of the dummy model are analyzed, and the capabilities of the dimension reduction method applicable to the high dimensional data are clarified. Then the principle of autoencoder is analyzed and its applicability is explained. Finally, the linear and nonlinear dimension reduction of the autoencoder is verified by using the dummy response data. The results show that the mean square error values of the linear and nonlinear dimension reduction reconstruction are 3.2% and 1.6% respectively, and the response curves obtained by the reconstruction are close to the original curves. The applicability and effectiveness of autoencoder for dimension reduction of dummy response data are proved.