python - What is the network size limit in PyBrain? -

i have constructed neural network in pybrain, ran in problem of network being big.

• so exact limits in pybrain?

• how can construct network 750000 inputs, 3 outputs , working ammount of hidden nodes(currently works 35-40 hidden nodes, feel lot little)?

• are there other ways construct image recognition(in case basic, not identical(random solid colors , sizes) shape recognition - square, circle , equilateral triangle) using mlp n input nodes n = image.width * image.height * 3?

with many nodes if there no computation errors amount of time network need perform single training epoch enormous. don't believe there specific limit network size in pybrain other common sense , computer capabilities.

also idea of neural network simplify complex problems. have shrink input data. , mean shrink lot, purpose 100 input nodes more enough.

i recommend source material choosing input data, , network architecture:

• localized precipitation forecasts numerical weather prediction model using artificial neural networks

• great answer network architecture

• general article on wikipedia

it's bit matter of experience chose appropriate input data, it.

hope helpful.