def bgan_synth(synth_dataset,
z_dim,
batch_size=64,
numz=5,
num_iter=1000,
wasserstein=False,
rpath="synth_results",
gen_observed=10,
base_learning_rate=1e-2,
lr_decay=3.,
save_weights=False):
bgan = BGAN(
[synth_dataset.x_dim],
z_dim,
synth_dataset.N,
batch_size=batch_size,
prior_std=10.0,
alpha=1e-3,
J=numz,
M=1,
ml=(numz == 1),
num_classes=1,
wasserstein=wasserstein, # unsupervised only
gen_observed=gen_observed)
print "Starting session"
session = get_session()
tf.global_variables_initializer().run()
print "Starting training loop"
num_train_iter = num_iter
sample_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
all_aics_fake, all_data_fake, all_dists = [], [], []
for train_iter in range(num_train_iter):
learning_rate = base_learning_rate * np.exp(-lr_decay * min(
1.0, (train_iter * batch_size) / float(synth_dataset.N)))
print learning_rate
batch_z = np.random.uniform(-1, 1, [batch_size, z_dim])
input_batch = synth_dataset.next_batch(batch_size)
_, d_loss = session.run(
[bgan.d_optim, bgan.d_loss],
feed_dict={
bgan.inputs: input_batch,
bgan.z: batch_z,
bgan.d_learning_rate: learning_rate
})
if wasserstein:
session.run(bgan.clip_d, feed_dict={})
g_losses = []
for gi in xrange(bgan.num_gen):
# compute g_sample loss
batch_z = np.random.uniform(-1, 1, [batch_size, z_dim])
_, g_loss = session.run(
[bgan.g_optims[gi], bgan.generation["g_losses"][gi]],
feed_dict={
bgan.z: batch_z,
bgan.g_learning_rate: learning_rate
})
g_losses.append(g_loss)
print "Disc loss = %.2f, Gen loss = %s" % (d_loss, ", ".join(
["%.2f" % gl for gl in g_losses]))
if (train_iter + 1) % 100 == 0:
print "Disc loss = %.2f, Gen loss = %s" % (d_loss, ", ".join(
["%.2f" % gl for gl in g_losses]))
print "Running GMM on sampled data"
fake_data = []
for num_samples in xrange(10):
for gi in xrange(bgan.num_gen):
# collect sample
sample_z = np.random.uniform(-1,
1,
size=(batch_size, z_dim))
sampled_data = session.run(
bgan.generation["gen_samplers"][gi],
feed_dict={bgan.z: sample_z})
fake_data.append(sampled_data)
X_real = synth_dataset.X
X_sample = np.concatenate(fake_data)
all_data_fake.append(X_sample)
aics_fake = gmm_ms(X_sample)
print "Fake number of clusters (AIC estimate):", aics_fake.argmin()
dist, X_trans_real, X_trans_fake = analyze_div(X_real, X_sample)
print "JS div:", dist
fp = FigPrinter((1, 2))
xmin1 = np.min(X_trans_real[:, 0]) - 1.0
xmax1 = np.max(X_trans_real[:, 0]) + 1.0
xmin2 = np.min(X_trans_real[:, 1]) - 1.0
xmax2 = np.max(X_trans_real[:, 1]) + 1.0
fp.ax_arr[0].plot(X_trans_real[:, 0], X_trans_real[:, 1], '.r')
fp.ax_arr[0].set_xlim([xmin1, xmax1])
fp.ax_arr[0].set_ylim([xmin2, xmax2])
fp.ax_arr[1].plot(X_trans_fake[:, 0], X_trans_fake[:, 1], '.g')
fp.ax_arr[1].set_xlim([xmin1, xmax1])
fp.ax_arr[1].set_ylim([xmin2, xmax2])
fp.ax_arr[0].set_aspect('equal', adjustable='box')
fp.ax_arr[1].set_aspect('equal', adjustable='box')
fp.ax_arr[1].set_title("Iter %i" % (train_iter + 1))
fp.print_to_file(
os.path.join(
rpath,
"pca_distribution_%i_%i.png" % (numz, train_iter + 1)))
all_aics_fake.append(aics_fake)
all_dists.append(dist)
if save_weights:
var_dict = {}
for var in tf.trainable_variables():
var_dict[var.name] = session.run(var.name)
np.savez_compressed(
os.path.join(rpath, "weights_%i.npz" % train_iter),
**var_dict)
return {
"data_fake": all_data_fake,
"data_real": synth_dataset.X,
"z_dim": z_dim,
"numz": numz,
"num_iter": num_iter,
"divergences": all_dists,
"all_aics_fake": np.array(all_aics_fake)
}
def bgan_synth(synth_dataset,
z_dim,
batch_size=64,
numz=5,
num_iter=1000,
wasserstein=False,
rpath="synth_results",
base_learning_rate=1e-2,
lr_decay=3.,
save_weights=False):
bgan = BGAN(
[synth_dataset.x_dim],
z_dim,
synth_dataset.N,
batch_size=batch_size,
prior_std=10.0,
alpha=1e-3,
J=numz,
M=1,
ml=(numz == 1),
num_classes=1,
wasserstein=wasserstein, # unsupervised only
)
print("Starting session")
session = get_session()
tf.global_variables_initializer().run()
print("Starting training loop")
num_train_iter = num_iter
all_aics_fake, all_data_fake, all_dists = [], [], []
it_loss, it_acc, acc = [], [], []
mean = np.random.uniform(-1, 1, size=z_dim)
cov = np.random.uniform(-1, 1, size=(z_dim, z_dim))
cov = np.dot(cov, cov.transpose())
for train_iter in range(num_train_iter):
learning_rate = base_learning_rate * np.exp(-lr_decay * min(
1.0, (train_iter * batch_size) / float(synth_dataset.N)))
print(learning_rate)
batch_z = np.random.uniform(-1, 1, [batch_size, z_dim])
#batch_z = np.random.multivariate_normal(mean, cov, size=batch_size)
#batch_z = np.random.normal(0, 1.0, size=(batch_size, z_dim))
_, input_batch = synth_dataset.next_batch(batch_size)
_, d_loss = session.run(
[bgan.d_optim, bgan.d_loss],
feed_dict={
bgan.inputs: input_batch,
bgan.z: batch_z,
bgan.d_learning_rate: learning_rate
})
if wasserstein:
session.run(bgan.clip_d, feed_dict={})
g_losses = []
for gi in range(bgan.num_gen):
# compute g_sample loss
batch_z = np.random.uniform(-1, 1, [batch_size, z_dim])
#batch_z = np.random.multivariate_normal(mean, cov, size=batch_size)
#batch_z = np.random.normal(0, 1.0, size=(batch_size, z_dim))
_, g_loss, prior = session.run([
bgan.g_optims[gi], bgan.generation["g_losses"][gi],
bgan.generation["g_prior"][gi]
],
feed_dict={
bgan.z: batch_z,
bgan.g_learning_rate:
learning_rate
})
print(prior, gi)
g_losses.append(g_loss)
print("Disc loss = %.2f, Gen loss = %s" %
(d_loss, ", ".join(["%.2f" % gl for gl in g_losses])))
if (train_iter + 1) % 100 == 0:
print("Disc loss = %.2f, Gen loss = %s" %
(d_loss, ", ".join(["%.2f" % gl for gl in g_losses])))
print("Running GMM on sampled data")
fake_data = []
for num_samples in range(10):
for gi in range(bgan.num_gen):
# collect sample
sample_z = np.random.uniform(-1,
1,
size=(batch_size, z_dim))
#sample_z = np.random.multivariate_normal(mean, cov, size=batch_size)
#sample_z = np.random.normal(0, 1.0, size=(batch_size, z_dim))
sampled_data = session.run(
bgan.generation["gen_samplers"][gi],
feed_dict={bgan.z: sample_z})
fake_data.append(sampled_data)
X_real = synth_dataset.X
X_sample = np.concatenate(fake_data)
all_data_fake.append(X_sample)
"""aics_fake = gmm_ms(X_sample)
print ("Fake number of clusters (AIC estimate):", aics_fake.argmin())"""
dist, X_trans_real, X_trans_fake = analyze_div(X_real, X_sample)
print("JS div:", dist)
"""fp = FigPrinter((1,2))
xmin1 = np.min(X_trans_real[:, 0]) - 1.0
xmax1 = np.max(X_trans_real[:, 0]) + 1.0
xmin2 = np.min(X_trans_real[:, 1]) - 1.0
xmax2 = np.max(X_trans_real[:, 1]) + 1.0
fp.ax_arr[0].plot(X_trans_real[:, 0], X_trans_real[:, 1], '.r')
fp.ax_arr[0].set_xlim([xmin1, xmax1]); fp.ax_arr[0].set_ylim([xmin2, xmax2])
fp.ax_arr[1].plot(X_trans_fake[:, 0], X_trans_fake[:, 1], '.g')
fp.ax_arr[1].set_xlim([xmin1, xmax1]); fp.ax_arr[1].set_ylim([xmin2, xmax2])
fp.ax_arr[0].set_aspect('equal', adjustable='box')
fp.ax_arr[1].set_aspect('equal', adjustable='box')
fp.ax_arr[1].set_title("Iter %i" % (train_iter+1))
fp.print_to_file(os.path.join(rpath, "pca_distribution_%i_%i.png" % (numz, train_iter+1)))"""
all_dists.append(dist)
it_loss.append(train_iter + 1)
if save_weights:
var_dict = {}
for var in tf.trainable_variables():
var_dict[var.name] = session.run(var.name)
np.savez_compressed(
os.path.join(rpath, "weights_%i.npz" % train_iter),
**var_dict)
if (train_iter + 1) % 200 == 0:
count_pos = 0
count = 0
for i in range(int(synth_dataset.N_test / batch_size)):
d_logit = session.run(
[bgan.D],
feed_dict={
bgan.inputs:
synth_dataset.X_test[i * batch_size:(i + 1) *
batch_size]
})
d_logit = d_logit[0]
count_pos += np.sum(np.argmax(d_logit, 1))
count += batch_size
acc.append(100.0 * count_pos / count)
it_acc.append(train_iter + 1)
fp = FigPrinter((1, 2))
fp.ax_arr[0].plot(it_loss, all_dists)
fp.ax_arr[0].set_title("JS Divergence")
fp.ax_arr[1].plot(it_acc, acc)
fp.ax_arr[1].set_title("Test Accuracy")
fp.print_to_file(os.path.join(rpath, "divergence_disc.png"))
return {
"data_fake": all_data_fake,
"data_real": synth_dataset.X,
"z_dim": z_dim,
"numz": numz,
"num_iter": num_iter,
"divergences": all_dists,
"it_loss": it_loss,
"acc": acc,
"it_acc": it_acc
}