How to change the activation function in Output Layer?

Hello, does any of you have any idea of how to manually implement the type of activation function for output layer: "OutputLayerActivation" for Regressor Neural Network in Statistical and Machine Learning Toolbox? I attempt to include it in fitrnet(), but it does not work. Thank you.

Answer 1

[Krishna]

Hello Cuthbert,

From what I gather, you are looking to manually implement an activation function and assign this custom function to "OutputLayerActivation". Unfortunately, when using the fitrnet function, you are limited to predefined activation functions such as 'relu', 'sigmoid', 'none', and 'tanh', as specified in the documentation:

https://www.mathworks.com/help/stats/fitrnet.html

However, you do have the option to utilize dlnetworks, which allows you to create your own layers that can serve as custom activation functions.

For more information on how to define custom layers, please review the following documentation:

https://www.mathworks.com/help/deeplearning/ug/define-custom-deep-learning-layers.html

Hope this helps.