Can somebody help me explain multilayer neural networks?

Question

I programming a neural network with 1 hiddel layer, but when I need more hidden layer, my program is not be able to use it.

I read some paper about multi hidden layer, and these training methods(example backpropagation), but I dont understand a few thing, I really want to understand, or someone show me an understable code. When I try to programming multilayer training, and prediction, something not working, so I would be grateful if someone link here a code example for multi hidden layer neural network
(I written with java)

Answer 1

That's fine. It's for the sake of easy understanding to break stuff into modules.

Answer 2

Did you model classes properly ? We have a Simple Cell class, with Input, Output class, Layer class - map / array of Simple Cells, Network class - List of Layer objects.

Answer 3

This is more of an LLD problem than of machine learning.

Answer 4

Google my man, just google - neural network implementation from scratch in java.

Answer 5

Use multiple threads to run different nodes on different cores. Nvidia does it very nice on GPUs.

Answer 6

I learnt in university and I did Andrew Ng free course around 3 years ago.

Answer 7

there is website called machinelearning.com i guess

Answer 8

Cool, have fun.

Answer 9

Did you learn about integration & differentiation in math ? Think of hidden layer's input as new neural network and try to recursively build logic.

Answer 10

I am not working multiple classes, I have 1 class, and the layers weights I store in matrices

Answer 11

Can you souggest me, sites, or app where I can see multiple layer's property, and arhitecture?

Answer 12

Thanks, and what can I do for optimalize a big neural network example with 10000 hidden nodes?

Answer 13

Thanks, the tips, where are you learn about neural networks?

Answer 14

Thanks the helping😊

Answer 15

I'd like to find a site, or any helping paper, about the backpropagation algorithm, because I think, I don't understand few thing in the method

In single hidden layer network I understand then traiming, and the error calculation, but more complicatin in multilayer network

Answer 16

I did not learn about integration, and....., but I know about these math things(so I cant application these math things for a problem), do you suggest me, to learn more math?

Answer 17

yes, just the basics are enough