def dimension_reduction(*x, algo='pca', **kwargs):
algo = str(algo).lower()
assert algo in ('pca', 'tsne', 'umap'), \
"No support for algorithm: '%s'" % algo
if x[0].shape[1] == 1:
raise ValueError("No dimension reduction for input with shape: %s" %
str(x[0].shape))
elif x[0].shape[1] == 2:
pass
elif algo == 'tsne':
x = fast_tsne(*x,
n_components=2,
perplexity=30.0,
learning_rate=200,
n_iter=1000,
random_state=1234,
n_jobs=8,
**kwargs)
elif algo == 'pca':
x = fast_pca(*x, n_components=2, random_state=1234, **kwargs)
else:
x = fast_umap(*x, random_state=1234, **kwargs)
if len(x) == 1:
return x[0]
return x
def extract_pca(p_train, p_test):
# p_train, p_test : the output and latent distributions
pca = [
fast_pca(squeeze(train.mean()), squeeze(test.mean()),
n_components=2)[-1] for train, test in zip(p_train, p_test)
if train.event_shape[0] > 1
]
return pca
_ = K.eval(loss, feed_dict={X: X_train[start:end]},
update_after=update_ops)
prog.add(end - start)
train_losses.append(_)
# ====== training log ====== #
print(ctext("[Epoch %d]" % epoch, 'yellow'), '%.2f(s)' % (timeit.default_timer() - start_time))
print("[Training set] Loss: %.4f" % np.mean(train_losses))
# ====== validation set ====== #
code_samples, lo = K.eval([Z, loss], feed_dict={X: X_valid})
print("[Valid set] Loss: %.4f" % lo)
# ====== record the history ====== #
record_train_loss.append(np.mean(train_losses))
record_valid_loss.append(lo)
# ====== plotting ====== #
if args.dim > 2:
code_samples = ml.fast_pca(code_samples, n_components=2,
random_state=K.get_rng().randint(10e8))
img_samples = f_samples()
img_mean = f_X(X_valid[:25])
V.plot_figure(nrow=3, ncol=12)
ax = plt.subplot(1, 3, 1)
ax.scatter(code_samples[:, 0], code_samples[:, 1], s=2, c=y_valid, alpha=0.3)
ax.set_title('Epoch %d' % epoch)
ax.set_aspect('equal', 'box')
ax.axis('off')
ax = plt.subplot(1, 3, 2)
ax.imshow(V.tile_raster_images(img_samples), cmap=plt.cm.Greys_r)
ax.axis('off')
_ = K.eval(loss, feed_dict={X: X_train[start:end]},
update_after=update_ops)
prog.add(end - start)
train_losses.append(_)
# ====== training log ====== #
print(ctext("[Epoch %d]" % epoch, 'yellow'), '%.2f(s)' % (timeit.default_timer() - start_time))
print("[Training set] Loss: %.4f" % np.mean(train_losses))
# ====== validation set ====== #
code_samples, lo = K.eval([Z, loss], feed_dict={X: X_valid})
print("[Valid set] Loss: %.4f" % lo)
# ====== record the history ====== #
record_train_loss.append(np.mean(train_losses))
record_valid_loss.append(lo)
# ====== plotting ====== #
if args.dim > 2:
code_samples = ml.fast_pca(code_samples, n_components=2,
random_state=K.get_rng().randint(10e8))
img_samples = f_samples()
img_mean = f_X(X_valid[:25])
V.plot_figure(nrow=3, ncol=12)
ax = plt.subplot(1, 3, 1)
ax.scatter(code_samples[:, 0], code_samples[:, 1], s=2, c=y_valid, alpha=0.3)
ax.set_title('Epoch %d' % epoch)
ax.set_aspect('equal', 'box')
ax.axis('off')
ax = plt.subplot(1, 3, 2)
ax.imshow(V.tile_raster_images(img_samples), cmap=plt.cm.Greys_r)
ax.axis('off')
random_state=5218)
scorer.fit(X=X_train, y=y_true['train'])
scorer.evaluate(X_test, y_true['test'], labels=labels)
# ====== svm scoring ====== #
print(ctext("==== '%s'" % "Ivec SVM-scoring", 'cyan'))
scorer = ml.Scorer(wccn=True, lda=True, method='svm')
scorer.fit(X=X_train, y=y_true['train'])
scorer.evaluate(X_test, y_true['test'], labels=labels)
# ===========================================================================
# Super-vector
# ===========================================================================
X_train = stats['train'][1]
X_test = stats['test'][1]
X_train, X_test = ml.fast_pca(X_train,
X_test,
n_components=args.tdim,
algo='ppca',
random_state=5218)
# ====== GMM scoring ====== #
print(ctext("==== '%s'" % "Super-Vector GMM-scoring-ova", 'cyan'))
scorer = ml.GMMclassifier(strategy="ova",
n_components=3,
covariance_type='full',
centering=True,
wccn=True,
unit_length=True,
lda=False,
concat=False)
scorer.fit(X=X_train, y=y_true['train'])
scorer.evaluate(X_test, y_true['test'], labels=labels)
# ====== plda scoring ====== #
from odin import ml from odin import visual as vs os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3' os.environ['TF_FORCE_GPU_ALLOW_GROWTH'] = 'true' tf.random.set_seed(8) np.random.seed(8) X, y = load_digits(return_X_y=True) X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3) X_umap = ml.fast_umap(X_train, X_test) X_tsne = ml.fast_tsne(X_train, X_test) X_pca = ml.fast_pca(X_train, X_test, n_components=2) styles = dict(size=12, alpha=0.6, centroids=True) vs.plot_figure(6, 12) vs.plot_scatter(x=X_pca[0], color=y_train, ax=(1, 2, 1), **styles) vs.plot_scatter(x=X_pca[1], color=y_test, ax=(1, 2, 2), **styles) vs.plot_figure(6, 12) vs.plot_scatter(x=X_tsne[0], color=y_train, ax=(1, 2, 1), **styles) vs.plot_scatter(x=X_tsne[1], color=y_test, ax=(1, 2, 2), **styles) vs.plot_figure(6, 12) vs.plot_scatter(x=X_umap[0], color=y_train, ax=(1, 2, 1), **styles) vs.plot_scatter(x=X_umap[1], color=y_test, ax=(1, 2, 2), **styles)
def plot_epoch(task):
if task is None:
curr_epoch = 0
else:
curr_epoch = task.curr_epoch
if not (curr_epoch < 5 or curr_epoch % 5 == 0):
return
rand = np.random.RandomState(seed=1234)
X, y = X_test, y_test
n_data = X.shape[0]
Z = f_z(X)
W, W_stdev_mcmc, W_stdev_analytic = f_w(X)
X_pca, W_pca_1 = fast_pca(X,
W,
n_components=2,
random_state=rand.randint(10e8))
W_pca_2 = fast_pca(W, n_components=2, random_state=rand.randint(10e8))
X_count_sum = np.sum(X, axis=tuple(range(1, X.ndim)))
W_count_sum = np.sum(W, axis=-1)
n_visual_samples = 8
nrow = 13 + n_visual_samples * 3
V.plot_figure(nrow=int(nrow * 1.8), ncol=18)
with V.plot_gridSpec(nrow=nrow + 3, ncol=6, hspace=0.8) as grid:
# plot the latent space
for i, (z, name) in enumerate(zip(Z, Z_names)):
if z.shape[1] > 2:
z = fast_pca(z,
n_components=2,
random_state=rand.randint(10e8))
ax = V.subplot(grid[:3, (i * 2):(i * 2 + 2)])
V.plot_scatter(x=z[:, 0],
y=z[:, 1],
color=y,
marker=y,
n_samples=4000,
ax=ax,
legend_enable=False,
legend_ncol=n_classes)
ax.set_title(name, fontsize=12)
# plot the reconstruction
for i, (x, name) in enumerate(
zip([X_pca, W_pca_1, W_pca_2], [
'Original data', 'Reconstruction',
'Reconstruction (separated PCA)'
])):
ax = V.subplot(grid[3:6, (i * 2):(i * 2 + 2)])
V.plot_scatter(x=x[:, 0],
y=x[:, 1],
color=y,
marker=y,
n_samples=4000,
ax=ax,
legend_enable=i == 1,
legend_ncol=n_classes,
title=name)
# plot the reconstruction count sum
for i, (x, count_sum, name) in enumerate(
zip([X_pca, W_pca_1], [X_count_sum, W_count_sum], [
'Original data (Count-sum)', 'Reconstruction (Count-sum)'
])):
ax = V.subplot(grid[6:10, (i * 3):(i * 3 + 3)])
V.plot_scatter(x=x[:, 0],
y=x[:, 1],
val=count_sum,
n_samples=2000,
marker=y,
ax=ax,
size=8,
legend_enable=i == 0,
legend_ncol=n_classes,
title=name,
colorbar=True,
fontsize=10)
# plot the count-sum series
count_sum_observed = np.sum(X, axis=0).ravel()
count_sum_expected = np.sum(W, axis=0)
count_sum_stdev_explained = np.sum(W_stdev_mcmc, axis=0)
count_sum_stdev_total = np.sum(W_stdev_analytic, axis=0)
for i, kws in enumerate([
dict(xscale='linear', yscale='linear', sort_by=None),
dict(xscale='linear', yscale='linear', sort_by='expected'),
dict(xscale='log', yscale='log', sort_by='expected')
]):
ax = V.subplot(grid[10:10 + 3, (i * 2):(i * 2 + 2)])
V.plot_series_statistics(count_sum_observed,
count_sum_expected,
explained_stdev=count_sum_stdev_explained,
total_stdev=count_sum_stdev_total,
fontsize=8,
title="Count-sum" if i == 0 else None,
**kws)
# plot the mean and variances
curr_grid_index = 13
ids = rand.permutation(n_data)
ids = ids[:n_visual_samples]
for i in ids:
observed, expected, stdev_explained, stdev_total = \
X[i], W[i], W_stdev_mcmc[i], W_stdev_analytic[i]
observed = observed.ravel()
for j, kws in enumerate([
dict(xscale='linear', yscale='linear', sort_by=None),
dict(xscale='linear', yscale='linear', sort_by='expected'),
dict(xscale='log', yscale='log', sort_by='expected')
]):
ax = V.subplot(grid[curr_grid_index:curr_grid_index + 3,
(j * 2):(j * 2 + 2)])
V.plot_series_statistics(observed,
expected,
explained_stdev=stdev_explained,
total_stdev=stdev_total,
fontsize=8,
title="Test Sample #%d" %
i if j == 0 else None,
**kws)
curr_grid_index += 3
V.plot_save(os.path.join(FIGURE_PATH, 'latent_%d.png' % curr_epoch),
dpi=200,
log=True)
exit()
def plot_imputation_scatter(self,
test=True,
pca=False,
color_by_library=True):
start_time = time.time()
n_system = len(self) + 2 # add the original and the corrupted
data_type = 'test' if test else 'train'
if n_system <= 5:
nrow = 1
ncol = n_system
else:
nrow = 2
ncol = int(np.ceil(n_system / 2))
X_org = self.posteriors[0].X_test_org if test else self.posteriors[
0].X_train_org
X_crr = self.posteriors[0].X_test if test else self.posteriors[
0].X_train
y = self.posteriors[0].y_test if test else self.posteriors[0].y_train
labels = self.posteriors[0].labels
is_binary_classes = self.posteriors[0].is_binary_classes
allV = [X_org, X_crr] + [
pos.V_test if test else pos.V_train for pos in self.posteriors
]
assert X_org.shape == X_crr.shape and all(v.shape == X_org.shape
for v in allV)
all_names = ["[%s]Original" % data_type,
"[%s]Corrupted" % data_type
] + [i.short_id_lines for i in self.posteriors]
# log-normalize everything
if len(X_org) > 5000:
np.random.seed(5218)
ids = np.random.permutation(X_org.shape[0])[:5000]
allV = [v[ids] for v in allV]
y = y[ids]
if is_binary_classes:
y = np.argmax(y, axis=-1)
else:
y = ProbabilisticEmbedding().fit_transform(y)
y = np.argmax(y, axis=-1)
allV = [log_norm(v) for v in allV]
fig = plt.figure(figsize=(min(20, 5 * ncol) + 2, nrow * 5))
for idx, (name, v) in enumerate(zip(all_names, allV)):
ax = plt.subplot(nrow, ncol, idx + 1)
n = np.sum(v, axis=-1)
v = fast_pca(v, n_components=2) if pca else fast_tsne(
v, n_components=2)
with catch_warnings_ignore(Warning):
if color_by_library:
plot_scatter(x=v,
val=n,
ax=ax,
size=8,
legend_enable=False,
grid=False,
title=name)
else:
plot_scatter(x=v,
color=[labels[i] for i in y],
marker=[labels[i] for i in y],
ax=ax,
size=8,
legend_enable=True if idx == 0 else False,
grid=False,
title=name)
with catch_warnings_ignore(Warning):
plt.tight_layout()
self.add_figure(
'imputation_scatter_%s_%s' %
('lib' if color_by_library else 'cell', data_type), fig)
return self._log(
'plot_imputation_scatter[%s] %s(s)' %
(data_type, ctext(time.time() - start_time, 'lightyellow')))
scorer = ml.PLDA(n_phi=TV_DIM // 2, n_iter=12,
centering=True, wccn=True, unit_length=True,
random_state=5218)
scorer.fit(X=X_train, y=y_true['train'])
scorer.evaluate(X_test, y_true['test'], labels=labels)
# ====== svm scoring ====== #
print(ctext("==== '%s'" % "Ivec SVM-scoring", 'cyan'))
scorer = ml.Scorer(wccn=True, lda=True, method='svm')
scorer.fit(X=X_train, y=y_true['train'])
scorer.evaluate(X_test, y_true['test'], labels=labels)
# ===========================================================================
# Super-vector
# ===========================================================================
X_train = stats['train'][1]
X_test = stats['test'][1]
X_train, X_test = ml.fast_pca(X_train, X_test, n_components=args.tdim,
algo='ppca', random_state=5218)
# ====== GMM scoring ====== #
print(ctext("==== '%s'" % "Super-Vector GMM-scoring-ova", 'cyan'))
scorer = ml.GMMclassifier(strategy="ova",
n_components=3, covariance_type='full',
centering=True, wccn=True, unit_length=True,
lda=False, concat=False)
scorer.fit(X=X_train, y=y_true['train'])
scorer.evaluate(X_test, y_true['test'], labels=labels)
# ====== plda scoring ====== #
print(ctext("==== '%s'" % "Super-Vector PLDA-scoring", 'cyan'))
scorer = ml.PLDA(n_phi=TV_DIM // 2, n_iter=12,
centering=True, wccn=True, unit_length=True,
random_state=5218)
scorer.fit(X=X_train, y=y_true['train'])
scorer.evaluate(X_test, y_true['test'], labels=labels)