def plot_validation_curves(key,keyRange):
print "Varying parameter: ",key
bValLoss = float('inf')
valLoss = []
trLoss = []
bidx = 0
for idx,v in enumerate(keyRange):
print "Testing parameter = ",v
param[key] = v
nnet = nn.nnet(d,k,param)
tl,vl = nnet.train(Xtr,Ytr,Xval,Yval)
valLoss.append(vl)
trLoss.append(tl)
if vl < bValLoss:
bValLoss = vl
bidx = idx
print "Best Validation Loss: ",bValLoss
print "Best Value for parameter ",key,": ",keyRange[bidx]
plt.plot(keyRange,valLoss,label='Validation Loss')
plt.plot(keyRange,trLoss,label='Training Loss')
plt.xlabel(key)
plt.ylabel("Cross-entropy Loss")
plt.title("Effects of NN Parameters on Cross-Entropy Loss")
plt.legend(loc='upper right')
plt.show()
return bidx
def setUp(self): self.inputDim = 4 self.numHidden = 3 self.outputDim = 1 self.data = np.array([ [0.1,0.2,0.9,0.7], [0,11,0.21,0.91,0.71], [0.12,0.22,0.92,0.72], [0.9,0.7,0.1,0.2], [0.91,0.71,0.11,0.21], [0.92,0.72,0.12,0.22] ]) self.targets = np.array([0],[0],[0],[1],[1],[1]) self.nnet = nn.nnet(inputDim,numHidden,outputDim)
Xtr = X[:,idx[0]] Ytr = Y[:,idx[0]] Xval = X[:,idx[1]] Yval = Y[:,idx[1]] Xte = X[:,idx[2]] Yte = Y[:,idx[2]] d = Xtr.shape[0] k = Yte.shape[0] # Vary number of hidden units key = 'nHid' keyRange = range(5,35,5) bidx = plot_validation_curves(key,keyRange) # Vary the learning rate # key = 'lrate' # keyRange = np.arange(0.1,1.5,0.1) # bidx = plot_validation_curves(key,keyRange) # Vary the momentum term # key = 'alpha' # keyRange = np.arange(0.1,1.0,0.1) # bidx = plot_validation_curves(key,keyRange) # Report error param[key] = keyRange[bidx] nnet = nn.nnet(d,k,param) nnet.train(Xtr,Ytr,Xval,Yval) pred,mce = nnet.predict(Xte,Yte) print "Misclassification error: ",mce
import numpy as np import neuralnet as nn nnet = nn.nnet(3,2) nnet.initialize_weights() W1 = nnet.W_in2hid W2 = nnet.W_hid2out X = np.array([[1,2,3],[4,5,6],[7,8,9]]) y = np.array([[0,1],[1,0],[1,0]]).T nnet.train(X,y)