Neural and RELu

[msynq...@gmail.com]

Hi there,

Please can you help me, I think I'm lost...

I have tried to implement the Accord.net DeepBeeliefNetwork sample but my network always returns the same output... and when I tune it with backpropagation it does not progress (always same error rate and always same output).
I've read a lot of papers. Some says that best results are with approximative 2 hidden layers (500, 150), some say that we have better results with (1000, 1000).
All these papers works with the MNIST dataset...

My data has approximatively as large input length as MNIST data (wich is 768 inputs pixels) so I thinked I'de better implement approximatively the same etwork but... No results !

My data is also more "complicated" since I have input1 which has a range of -1:1, input2 has a range of 0:800000,... and so on (this is why I normalized it)

So, I decided to start with Neural Network and Relu but... I can't find samples with Accord.net

Does anyone can give me a simple code sample with Accord.net implementing Relu with deep Network ?!

I precise that I've normalized my data (with 0 mean and derivative but I'm not sure if I had to apply these 2).

And I precise that I'm a bit new to Neural networks.