def inf_train_gen(DATASET, BATCH_SIZE):
if DATASET == '25gaussians':
dataset = []
for i in range(int(100000 / 25)):
for x in range(-2, 3):
for y in range(-2, 3):
point = np.random.randn(2) * 0.05
point[0] += 2 * x
point[1] += 2 * y
dataset.append(point)
dataset = np.array(dataset, dtype='float32')
np.random.shuffle(dataset)
dataset /= 2.828 # stdev
while True:
for i in range(int(len(dataset) / BATCH_SIZE)):
yield dataset[i * BATCH_SIZE:(i + 1) * BATCH_SIZE]
elif DATASET == 'stacked_mnist':
ds = Reader.DS('stacked_train.npy')
np.save('../dataset/Stacked_MNIST/dist_info.npy', ds.labels)
while True:
# p = np.random.permutation(a.size)
# dd = ds.data[p]
# ll = ds.labels[p]
for i in range(int(ds.size / BATCH_SIZE)):
start = i * BATCH_SIZE
end = (i + 1) * BATCH_SIZE
yield ds.images[start:end], ds.labels[start:end]
elif DATASET == '1200D':
ds = np.load('../dataset/1200D/1200D_train.npy')
ds = ds.item()
ds_images = ds['images']
ds_dist = ds['y_dist']
means = ds['means']
ds_size = ds_images.shape[0]
np.save('../dataset/1200D/dist_ydist.npy', ds_dist)
np.save('../dataset/1200D/dist_means.npy', means)
while True:
for i in range(int(ds_size / BATCH_SIZE)):
start = i * BATCH_SIZE
end = (i + 1) * BATCH_SIZE
yield ds_images[start:end], -1
elif DATASET == 'swissroll':
while True:
data = datasets.make_swiss_roll(n_samples=BATCH_SIZE,
noise=0.25)[0]
data = data.astype('float32')[:, [0, 2]]
data /= 7.5 # stdev plus a little
yield data
elif DATASET == '8gaussians':
scale = 2.
centers = [(1, 0), (-1, 0), (0, 1), (0, -1),
(1. / np.sqrt(2), 1. / np.sqrt(2)),
(1. / np.sqrt(2), -1. / np.sqrt(2)),
(-1. / np.sqrt(2), 1. / np.sqrt(2)),
(-1. / np.sqrt(2), -1. / np.sqrt(2))]
centers = [(scale * x, scale * y) for x, y in centers]
while True:
dataset = []
for i in range(BATCH_SIZE):
point = np.random.randn(2) * .02
center = random.choice(centers)
point[0] += center[0]
point[1] += center[1]
dataset.append(point)
dataset = np.array(dataset, dtype='float32')
dataset /= 1.414 # stdev
yield dataset
