Activation function of Regression Neural Net in Azure ML Studio?

Question

I am not able to find activation function for Regression Neural Network in Azure Machine Learning Studio. I am not able to identify what is the activation function taken for my NN. Followed this document also-

https://docs.microsoft.com/en-us/azure/machine-learning/studio-module-reference/neural-network-regression

Can someone suggest where to mention it/ what is the default activation function used?

Answer 1

Microsoft Azure ML studio has basically two configurations to work:

1. Use the default architecture provided to create a neural network model (preferable for beginners).

   -In this you can only change the number of nodes in the hidden layer, learning rate, and normalization.

2. Defining your own custom architecture to create a neural network model (preferable if you know your way around neural networks).

   -In this you can customize the architecture completely and modify its connections along with the activation function of your choice.

Take a look at the following link for more detailed description (official documentation):

https://docs.microsoft.com/en-us/azure/machine-learning/studio-module-reference/neural-network-regression

The following example uses softmax activation:

https://gallery.azure.ai/Experiment/7d3f74981b5b42cd9687370671c86696

Default activation function is sigmoid for classification models and linear for regression models.