:: Doubt Multilayer Perceptron - Weka ::

[LeonardoPio]

Dear sir,

I am to use Time Series Analysis and Forecasting with Weka and the Algoritm Multilayer Perceptron and I have a fews doubts, can you help me?

1- How I change the number of nodes of hidden layer?

2 - The best params for I change is learningRate, hiddenLayers, momentum, epochs, validationThreshold? Can you recommend me a paper that explain this?

Best Regards,


Leo

[Mark]

In the editor dialog for the MultilayerPerceptron there is a button labeled "More" in the "About" field at the top - you can click this to get more information about each of the options. The description for the hiddenLayers option is:

"hiddenLayers -- This defines the hidden layers of the neural network. This is a list of positive whole numbers. 1 for each hidden layer. Comma seperated. To have no hidden layers put a single 0 here. This will only be used if autobuild is set. There are also wildcard values 'a' = (attribs + classes) / 2, 'i' = attribs, 'o' = classes , 't' = attribs + classes."

I'd suggest finding a good AI or machine learning book and reading about neural networks.

Cheers,
Mark.

[LeonardoPio]

Dear Mark,

Thanks for your answer,

I understand how I change the hidden layers, but in Weka I can change the nodes and weight? or only the hidden layers?

Cheers,

Leo