Neural network test: Integer classification

nn-int.py

from pybrain.tools.shortcuts import buildNetwork
from pybrain.structure import FeedForwardNetwork
from pybrain.datasets import SupervisedDataSet
from pybrain.supervised.trainers import BackpropTrainer
from pybrain.structure import LinearLayer, SigmoidLayer, TanhLayer
from pybrain.structure import FullConnection

import random

def int2bin(a):
	arr = [int(x) for x in bin(a)[2:]]
	return [0]*(4-len(arr)) + arr


# Network
net = FeedForwardNetwork()

# Layers
inLayer = LinearLayer(4)
hiddenLayer1 = SigmoidLayer(8)
hiddenLayer2 = SigmoidLayer(16)
hiddenLayer3 = SigmoidLayer(32)
hiddenLayer4 = SigmoidLayer(8)
outLayer = LinearLayer(1)

net.addInputModule(inLayer)
net.addModule(hiddenLayer1)
net.addModule(hiddenLayer2)
net.addModule(hiddenLayer3)
net.addModule(hiddenLayer4)
net.addOutputModule(outLayer)

# Connection
in_to_hidden1 = FullConnection(inLayer, hiddenLayer1)
hidden1_to_hidden2 = FullConnection(hiddenLayer1, hiddenLayer2)
hidden2_to_hidden3 = FullConnection(hiddenLayer2, hiddenLayer3)
hidden3_to_hidden4 = FullConnection(hiddenLayer3, hiddenLayer4)
hidden4_to_out = FullConnection(hiddenLayer4, outLayer)

net.addConnection(in_to_hidden1)
net.addConnection(hidden1_to_hidden2)
net.addConnection(hidden2_to_hidden3)
net.addConnection(hidden3_to_hidden4)
net.addConnection(hidden4_to_out)

# init network
net.sortModules()


# Data set
ds = SupervisedDataSet(4, 1)

#numbers = []
#for num in range(0,10):
#	numbers += [int(random.randint(0, 15))]
numbers = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,15]
for num in numbers:
	ds.addSample(tuple(int2bin(num)), (num,))
#print numbers
print ds


# Traning
trainer = BackpropTrainer(net, ds)
if True:
	i = 0
	while i < 100:
		err = trainer.train()
		print "Traning error:", err
		i=i+1
else:
	err = trainer.trainUntilConvergence()



# Activation
for i in range(16):
	print i, net.activate(int2bin(i))



#trainer.trainUntilConvergence()