def log_callback_em_conductance(embeddings, adjancy_matrix, n_centroid):
kmeans = pem.PoincareEM(n_centroid)
kmeans.fit(embeddings)
i = kmeans.predict(embeddings)
r = torch.arange(0, i.size(0), device=i.device)
prediction = torch.zeros(embeddings.size(0), n_centroid)
prediction[r, i] = 1
return {
"conductance": evaluation.mean_conductance(prediction, adjancy_matrix)
}
for i in range(50):
tloss = 0.
for x, y in dataloader:
optimizer.zero_grad()
pe_x = model(x.long())
pe_y = model(y.long())
ne = model((torch.rand(len(x), 10) * len(X)).long())
loss = graph_embedding_criterion(pe_x, pe_y, z=ne,
manifold=manifold).sum()
tloss += loss.item()
loss.backward()
optimizer.step()
print('Loss value for iteration ', i, ' is ', tloss)
em_alg = poincare_em.PoincareEM(13)
em_alg.fit(model.weight.data)
NF = em_alg.get_normalisation_coef()
pi, mu, sigma = em_alg.get_parameters()
pik = em_alg.get_pik(model.weight.data)
plot_poincare_disc_embeddings(model.weight.data.numpy(),
labels=dataset.Y,
save_folder="LOG/community_loss",
file_name="LFR_before_community_loss.png",
centroids=mu)
optimizer = rsgd.RSGD(model.parameters(), 2e-2, manifold=manifold)
dataset_o3 = corpora_tools.from_indexable(
# This script estimate gmm
import torch
from rcome.clustering_tools import poincare_em
from rcome.visualisation_tools.plot_tools import plot_poincare_gmm
# generating random set draw fro uniform
set_A = torch.randn(50, 2) * 2e-1 + 0.3
set_B = torch.randn(50, 2) * 1e-1 - 0.5
representation = torch.cat((set_A, set_B), 0)
out_of_disc = (representation[representation.norm(2, -1) > 1].norm(2, -1) +
1e-2)
representation[representation.norm(2,-1)>1] = \
torch.einsum('ij, i -> ij', representation[representation.norm(2,-1)>1], 1/out_of_disc)
# estimate the gmm
em_alg = poincare_em.PoincareEM(2)
em_alg.fit(representation)
Y = torch.zeros(100, 1).long()
Y[:50] = 1
plot_poincare_gmm(representation,
em_alg,
labels=Y,
save_folder="LOG/gmm_estimation",
file_name="gmm_estimation.png")