(ii) Interpretable neural network based on network structure constraints

In this chapter, we first test the performance of sosxnn model through simulation experiments. The results show that sosxnn model can decompose complex functions into several simple components, and achieve similar prediction accuracy with the classical black box algorithm. Then, we use the lending Club data set of American P2P lending company to show the application of s0sxn model in the financial field.

2.png

3.png

Experience it now and start the journey of digital transformation !