references = []
for i in range(network.model[layerToObserve]):
print "Sampling neuron " + str(i)
myimg = Image.new("L", (network.imageWidth, network.imageHeight), "white")
if samplesForAverage <= 1:
input = np.zeros(network.model[layerToObserve])
input[i] = 1
if samplesForAverage > 1:
input = np.zeros((samplesForAverage, network.model[layerToObserve]))
input[:, i] = np.ones(samplesForAverage)
neuronWeights = rbm.sample(input, layerToObserve, 0, binarize)
if samplesForAverage > 1:
neuronWeights = np.mean(neuronWeights, 0)
myimg.putdata(neuronWeights * 255)
myimg = myimg.transpose(Image.TRANSPOSE)
imagesPerLine = canvasWidth / network.imageWidth - 2
image = ImageTk.PhotoImage(myimg)
imagesprite = canvas.create_image(
20 + (network.imageWidth + 2) * (i % imagesPerLine),
20 + (network.imageHeight + 2) * (i / imagesPerLine),
image=image,
)
references = []
for i in range(network.model[layerToObserve]):
print "Sampling neuron " + str(i)
myimg = Image.new("L", (network.imageWidth, network.imageHeight), "white")
if (samplesForAverage <= 1):
input = np.zeros(network.model[layerToObserve])
input[i] = 1
if (samplesForAverage > 1):
input = np.zeros((samplesForAverage, network.model[layerToObserve]))
input[:, i] = np.ones(samplesForAverage)
neuronWeights = rbm.sample(input, layerToObserve, 0, binarize)
if (samplesForAverage > 1):
neuronWeights = np.mean(neuronWeights, 0)
myimg.putdata(neuronWeights * 255)
myimg = myimg.transpose(Image.TRANSPOSE)
imagesPerLine = canvasWidth / network.imageWidth - 2
image = ImageTk.PhotoImage(myimg)
imagesprite = canvas.create_image(
20 + (network.imageWidth + 2) * (i % imagesPerLine),
20 + (network.imageHeight + 2) * (i / imagesPerLine),
image=image)