How can I modify objective function in "trainingOptions" to suit a particular unusual problem

1 次查看（过去 30 天）
Kittipong Noonoi
Kittipong Noonoi 2018-10-30
回答： Greg Heath 2018-12-19
I'm trying to solve free-boundary problem with neural network and I find that Neural Network Toolbox is very helpful in constructing network but the problem is I don't know how to input my particular training function which is of the form (DE(LHS)+upper boundary +lower boundary)^2 instead of typical 'mse' e.g.(out-target)^2.

请先登录，再回答此问题。

回答（1 个）

Greg Heath
Greg Heath 2018-12-19
I think you are confused.
1. Your input and target define your function
2. In general, the standard net with tansig hidden nodes and linear output nodes will suffice.
3. However, if there is a physical or mathematical reason that the output has upper and/or lower bounds, then you can improve your solution by using logsig or tansig in the output layer
Hope this helps
Thank you for formally accepting my answer
Greg

类别

在 Help Center 和 File Exchange 中查找有关 Sequence and Numeric Feature Data Workflows 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by