Using DeepLearning Toolbox to approximate function - possible?

1 次查看（过去 30 天）
Carsten Daldrup
Carsten Daldrup 2020-10-8
评论： Mahesh Taparia 2020-10-19
Hey everyone,
I want to approximate a function f(x). I tried the NNTool and it worked - but not with the activation-function and transfer-function I want to use.
So I wanted to try the DeepLearning Toolbox. I created the network but I have problems with my data. I have just a simple function like f(x)=x^2 and the data x and f(x)=y. What kind of input-Layer I have to choose? And how do I have to prepare my data?
Regards

请先登录，再回答此问题。

回答（1 个）

Mahesh Taparia
Mahesh Taparia 2020-10-17
Hi
If the input output relation is known, then it straight forward to use that relation and estimate the result. In your case, it seems its straight forward.
Even if you want to use learning based approach, you can use the feedforwardnet. For more information, refer this documentation. Hope it will help.
  2 个评论
Carsten Daldrup
Carsten Daldrup 2020-10-19
Thanks for you answer, but that´s not what I was looking for. It works with DeepLearing and the new release 2020b. FeatureInput-Layer is the right choise.
Mahesh Taparia
Mahesh Taparia 2020-10-19
Hi
If this is the case, then you can use the fully connected layer followed by activations and the classification/ regression layer depending upon the task. You can refer here for an example.

请先登录，再进行评论。

类别

在 Help Center 和 File Exchange 中查找有关 Deep Learning Toolbox 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