def visualize_gibbs_result(self, start_x, gibbs_steps, filename):
# Run minibatch_size chains for gibbs_steps
x_samples = None
if not start_x is None:
x_samples = self.crbm.gibbs_samples_from(start_x, gibbs_steps)
else:
x_samples = self.crbm.random_gibbs_samples(gibbs_steps)
x_samples = x_samples.reshape((self.minibatch_size, 28*28))
tile = tile_raster_images(x_samples, self.image_size,
(1, self.minibatch_size), output_pixel_vals=True)
filepath = os.path.join(IMAGE_OUTPUT_DIR, filename+".png")
img = Image.fromarray(tile)
img.save(filepath)
print "Result of running Gibbs", \
gibbs_steps, "times outputed to", filepath
def visualize_filters(self, filename):
cp = self.cp
# filter size
fsz = (cp.height_filters, cp.width_filters)
tile_shape = (cp.num_filters, cp.num_input_planes)
filters_flattened = self.crbm.W.value.reshape(
(tile_shape[0]*tile_shape[1],
fsz[0]*fsz[1]))
tile = tile_raster_images(filters_flattened, fsz,
tile_shape, output_pixel_vals=True)
filepath = os.path.join(IMAGE_OUTPUT_DIR, filename+".png")
img = Image.fromarray(tile)
img.save(filepath)
print "Filters (as images) outputed to", filepath
def sample_from_rbm(self, gibbs_steps, test_set_x):
# find out the number of test samples
number_of_test_samples = test_set_x.value.shape[0]
# pick random test examples, with which to initialize the persistent chain
test_idx = rng.randint(number_of_test_samples - 20)
persistent_vis_chain = theano.shared(test_set_x.value[test_idx : test_idx + 20])
# define one step of Gibbs sampling (mf = mean-field)
[hid_mf, hid_sample, vis_mf, vis_sample] = self.rbm.gibbs_vhv(persistent_vis_chain)
# the sample at the end of the channel is returned by ``gibbs_1`` as
# its second output; note that this is computed as a binomial draw,
# therefore it is formed of ints (0 and 1) and therefore needs to
# be converted to the same dtype as ``persistent_vis_chain``
vis_sample = T.cast(vis_sample, dtype=theano.config.floatX)
# construct the function that implements our persistent chain
# we generate the "mean field" activations for plotting and the actual samples for
# reinitializing the state of our persistent chain
sample_fn = theano.function([], [vis_mf, vis_sample], updates={persistent_vis_chain: vis_sample})
# sample the RBM, plotting every `plot_every`-th sample; do this
# until you plot at least `n_samples`
n_samples = 10
plot_every = 1000
for idx in xrange(n_samples):
# do `plot_every` intermediate samplings of which we do not care
for jdx in xrange(plot_every):
vis_mf, vis_sample = sample_fn()
# construct image
image = PIL.Image.fromarray(
tile_raster_images(X=vis_mf, img_shape=(28, 28), tile_shape=(10, 10), tile_spacing=(1, 1))
)
image.save("sample_%i_step_%i.png" % (idx, idx * jdx))
