Effects of loss function and data sparsity on smooth manifold extraction with deep model

作者：

Highlights：

• Proper loss function selection improves smoothness better than the reduction of data sparsity.

• The less sensitive a deep model is to data sparsity, the smoother the extracted manifold is.

• Simply stacking hidden layers in deep model does not significantly improve smoothness.

摘要

•Proper loss function selection improves smoothness better than the reduction of data sparsity.•The less sensitive a deep model is to data sparsity, the smoother the extracted manifold is.•Simply stacking hidden layers in deep model does not significantly improve smoothness.

论文关键词：Smooth manifold,Deep model,Loss function,Data sparsity,Brenier theorem,Optimal Transport Theory

论文评审过程：Received 29 April 2020, Revised 25 February 2022, Accepted 7 March 2022, Available online 9 March 2022, Version of Record 17 March 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.116851