def kmean_poincare_ball():
n_samples = 20
dim = 2
n_clusters = 2
manifold = Hyperbolic(dimension=dim, point_type='ball')
metric = manifold.metric
cluster_1 = gs.random.uniform(low=0.5, high=0.6, size=(n_samples, dim))
cluster_2 = gs.random.uniform(low=0, high=-0.2, size=(n_samples, dim))
data = gs.concatenate((cluster_1, cluster_2), axis=0)
kmeans = RiemannianKMeans(riemannian_metric=metric,
n_clusters=n_clusters,
init='random',
mean_method='frechet-poincare-ball'
)
centroids = kmeans.fit(X=data, max_iter=100)
labels = kmeans.predict(X=data)
plt.figure(1)
colors = ['red', 'blue']
ax = visualization.plot(
data,
space='H2_poincare_disk',
marker='.',
color='black',
point_type=manifold.point_type)
for i in range(n_clusters):
ax = visualization.plot(
data[labels == i],
ax=ax,
space='H2_poincare_disk',
marker='.',
color=colors[i],
point_type=manifold.point_type)
ax = visualization.plot(
centroids,
ax=ax,
space='H2_poincare_disk',
marker='*',
color='green',
s=100,
point_type=manifold.point_type)
ax.set_title('Kmeans on Poincaré Ball Manifold')
return plt
def kmean_hypersphere():
"""Run K-means on the sphere."""
n_samples = 50
dim = 2
n_clusters = 2
manifold = Hypersphere(dim)
metric = manifold.metric
# Generate data on north pole
cluster_1 = manifold.random_von_mises_fisher(kappa=50, n_samples=n_samples)
# Generate data on south pole
cluster_2 = manifold.random_von_mises_fisher(kappa=50, n_samples=n_samples)
for point in cluster_2:
point[2] = -point[2]
data = gs.concatenate((cluster_1, cluster_2), axis=0)
kmeans = RiemannianKMeans(metric, n_clusters, tol=1e-3)
kmeans.fit(data)
labels = kmeans.predict(data)
centroids = kmeans.centroids
plt.figure(2)
colors = ['red', 'blue']
ax = visualization.plot(
data,
space='S2',
marker='.',
color='black')
for i in range(n_clusters):
if len(data[labels == i]) > 0:
ax = visualization.plot(
points=data[labels == i],
ax=ax,
space='S2',
marker='.',
color=colors[i])
ax = visualization.plot(
centroids,
ax=ax,
space='S2',
marker='*',
s=200,
color='green')
ax.set_title('Kmeans on Hypersphere Manifold')
return plt
def kmean_poincare_ball():
"""Run K-means on the Poincare ball."""
n_samples = 20
dim = 2
n_clusters = 2
manifold = PoincareBall(dim=dim)
metric = manifold.metric
cluster_1 = gs.random.uniform(low=0.5, high=0.6, size=(n_samples, dim))
cluster_2 = gs.random.uniform(low=0, high=-0.2, size=(n_samples, dim))
data = gs.concatenate((cluster_1, cluster_2), axis=0)
kmeans = RiemannianKMeans(metric=metric, n_clusters=n_clusters, init="random")
centroids = kmeans.fit(X=data)
labels = kmeans.predict(X=data)
plt.figure(1)
colors = ["red", "blue"]
ax = visualization.plot(
data,
space="H2_poincare_disk",
marker=".",
color="black",
point_type=manifold.point_type,
)
for i in range(n_clusters):
ax = visualization.plot(
data[labels == i],
ax=ax,
space="H2_poincare_disk",
marker=".",
color=colors[i],
point_type=manifold.point_type,
)
ax = visualization.plot(
centroids,
ax=ax,
space="H2_poincare_disk",
marker="*",
color="green",
s=100,
point_type=manifold.point_type,
)
ax.set_title("Kmeans on Poincaré Ball Manifold")
return plt
def test_hypersphere_kmeans_predict(self):
gs.random.seed(1234)
manifold = hypersphere.Hypersphere(2)
metric = hypersphere.HypersphereMetric(2)
x = manifold.random_von_mises_fisher(kappa=100, n_samples=200)
kmeans = RiemannianKMeans(metric, 5, tol=1e-5)
kmeans.fit(x, max_iter=100)
result = kmeans.predict(x)
centroids = kmeans.centroids
expected = gs.array([int(metric.closest_neighbor_index(x_i, centroids))
for x_i in x])
self.assertAllClose(expected, result)
def kmean_hypersphere():
"""Run K-means on the sphere."""
n_samples = 50
dim = 2
n_clusters = 2
manifold = Hypersphere(dim)
metric = manifold.metric
# Generate data on north pole
cluster_1 = manifold.random_von_mises_fisher(kappa=50, n_samples=n_samples)
# Generate data on south pole
cluster_2 = manifold.random_von_mises_fisher(kappa=50, n_samples=n_samples)
cluster_2 = -cluster_2
data = gs.concatenate((cluster_1, cluster_2), axis=0)
kmeans = RiemannianKMeans(metric, n_clusters, tol=1e-3)
kmeans.fit(data)
labels = kmeans.predict(data)
centroids = kmeans.centroids
plt.figure(2)
colors = ["red", "blue"]
ax = visualization.plot(data, space="S2", marker=".", color="black")
for i in range(n_clusters):
if len(data[labels == i]) > 0:
ax = visualization.plot(
points=data[labels == i], ax=ax, space="S2", marker=".", color=colors[i]
)
ax = visualization.plot(
centroids, ax=ax, space="S2", marker="*", s=200, color="green"
)
ax.set_title("Kmeans on the sphere")
return plt
def main():
cluster_1 = gs.random.uniform(low=0.5, high=0.6, size=(20, 2))
cluster_2 = gs.random.uniform(low=0, high=-0.2, size=(20, 2))
ax = plt.gca()
merged_clusters = gs.concatenate((cluster_1, cluster_2), axis=0)
manifold = Hyperbolic(dimension=2, point_type='ball')
metric = HyperbolicMetric(dimension=2, point_type='ball')
visualization.plot(
merged_clusters,
ax=ax,
space='H2_poincare_disk',
marker='.',
color='black',
point_type=manifold.point_type)
kmeans = RiemannianKMeans(
riemannian_metric=metric,
n_clusters=2,
init='random',
)
centroids = kmeans.fit(X=merged_clusters, max_iter=1)
labels = kmeans.predict(X=merged_clusters)
visualization.plot(
centroids,
ax=ax,
space='H2_poincare_disk',
marker='.',
color='red',
point_type=manifold.point_type)
print('Data_labels', labels)
plt.show()
def fit(self, data):
"""Fit a Gaussian mixture model (GMM) given the data.
Alternates between Expectation and Maximization steps
for some number of iterations.
Parameters
----------
data : array-like, shape=[n_samples, n_features]
Training data, where n_samples is the number of samples
and n_features is the number of features.
Returns
-------
self : object
Return the components of the computed
Gaussian mixture model: means, variances and mixture_coefficients.
"""
self._dimension = data.shape[-1]
if self.initialisation_method == 'kmeans':
kmeans = RiemannianKMeans(metric=self.metric,
n_clusters=self.n_gaussians,
init='random',
mean_method='batch',
lr=self.lr_mean)
centroids = kmeans.fit(X=data)
labels = kmeans.predict(X=data)
self.means = centroids
self.variances = gs.zeros(self.n_gaussians)
labeled_data = gs.vstack([labels, gs.transpose(data)])
labeled_data = gs.transpose(labeled_data)
for label, centroid in enumerate(centroids):
label_mask = gs.where(labeled_data[:, 0] == label)
grouped_by_label = labeled_data[label_mask][:, 1:]
v = variance(grouped_by_label, centroid, self.metric)
if grouped_by_label.shape[0] == 1:
v += MIN_VAR_INIT
self.variances[label] = v
else:
self.means = (gs.random.rand(self.n_gaussians, self._dimension) -
0.5) / self._dimension
self.variances = gs.random.rand(self.n_gaussians) / 10 + 0.8
self.mixture_coefficients = \
gs.ones(self.n_gaussians) / self.n_gaussians
posterior_probabilities = gs.ones((data.shape[0], self.means.shape[0]))
self.variances_range,\
self.normalization_factor_var, \
self.phi_inv_var =\
self.normalization_factor_init(
gs.arange(
ZETA_LOWER_BOUND, ZETA_UPPER_BOUND, ZETA_STEP))
for epoch in range(self.max_iter):
old_posterior_probabilities = posterior_probabilities
posterior_probabilities = self._expectation(data)
condition = gs.mean(
gs.abs(old_posterior_probabilities - posterior_probabilities))
if condition < EM_CONV_RATE and epoch > MINIMUM_EPOCHS:
logging.info('EM converged in %s iterations', epoch)
return self.means, self.variances, self.mixture_coefficients
self._maximization(data, posterior_probabilities)
logging.info('WARNING: EM did not converge \n'
'Please increase MINIMUM_EPOCHS.')
return self.means, self.variances, self.mixture_coefficients
def main():
"""Learning Poincaré graph embedding.
Learns Poincaré Ball embedding by using Riemannian
gradient descent algorithm. Then K-means is applied
to learn labels of each data sample.
"""
gs.random.seed(1234)
karate_graph = load_karate_graph()
hyperbolic_embedding = HyperbolicEmbedding()
embeddings = hyperbolic_embedding.embed(karate_graph)
colors = {1: 'b', 2: 'r'}
group_1 = mpatches.Patch(color=colors[1], label='Group 1')
group_2 = mpatches.Patch(color=colors[2], label='Group 2')
circle = visualization.PoincareDisk(point_type='ball')
_, ax = plt.subplots(figsize=(8, 8))
ax.axes.xaxis.set_visible(False)
ax.axes.yaxis.set_visible(False)
circle.set_ax(ax)
circle.draw(ax=ax)
for i_embedding, embedding in enumerate(embeddings):
x = embedding[0]
y = embedding[1]
pt_id = i_embedding
plt.scatter(
x, y,
c=colors[karate_graph.labels[pt_id][0]],
s=150
)
ax.annotate(pt_id, (x, y))
plt.tick_params(
which='both')
plt.title('Poincare Ball Embedding of the Karate Club Network')
plt.legend(handles=[group_1, group_2])
plt.show()
n_clusters = 2
kmeans = RiemannianKMeans(
riemannian_metric=hyperbolic_embedding.manifold.metric,
n_clusters=n_clusters,
init='random',
mean_method='frechet-poincare-ball')
centroids = kmeans.fit(X=embeddings, max_iter=100)
labels = kmeans.predict(X=embeddings)
colors = ['g', 'c', 'm']
circle = visualization.PoincareDisk(point_type='ball')
_, ax2 = plt.subplots(figsize=(8, 8))
circle.set_ax(ax2)
circle.draw(ax=ax2)
ax2.axes.xaxis.set_visible(False)
ax2.axes.yaxis.set_visible(False)
group_1_predicted = mpatches.Patch(
color=colors[0], label='Predicted Group 1')
group_2_predicted = mpatches.Patch(
color=colors[1], label='Predicted Group 2')
group_centroids = mpatches.Patch(
color=colors[2], label='Cluster centroids')
for _ in range(n_clusters):
for i_embedding, embedding in enumerate(embeddings):
x = embedding[0]
y = embedding[1]
pt_id = i_embedding
if labels[i_embedding] == 0:
color = colors[0]
else:
color = colors[1]
plt.scatter(
x, y,
c=color,
s=150
)
ax2.annotate(pt_id, (x, y))
for _, centroid in enumerate(centroids):
x = centroid[0]
y = centroid[1]
plt.scatter(
x, y,
c=colors[2],
marker='*',
s=150,
)
plt.title('K-means applied to Karate club embedding')
plt.legend(handles=[group_1_predicted, group_2_predicted, group_centroids])
plt.show()