How to find the optimum number of neurons and hidden layers for ANN?

How to find the optimum number of neurons and hidden layers for ANN?

Answer 1

[Matt J]

By trial and error, though you can automate some of that using the Experiment Manager.

Answer 2

[Ganesh]

Hi @Sunita,

I understand you are trying to train your ANN and would like to know the ideal number of neurons and hidden layers you need to use.

In Deep Learning, there is no "ideal" number for either of those parameters. The number of neurons on the input layers is equal to the number of Features in your dataset and the number of neurons on the output layer should be the number of Labels in your dataset. The other layers and neurons depends on the compelixity of your dataset.

If your dataset is small enough, you can try to automate the training by implementing different number of neurons and hidden layers. In some cases, the number of neurons follow the following number:

Hope this helps!