def test_aligned_update_array_error(aligned_iris, aligned_iris_relations):
data, target = aligned_iris
n_neighbors = [15, 15, 15, 15, 15]
small_aligned_model = AlignedUMAP(n_neighbors=n_neighbors[:3])
small_aligned_model.fit(data[:3], relations=aligned_iris_relations[:2])
with pytest.raises(ValueError):
small_aligned_model.update(data[3:],
relations=aligned_iris_relations[2:],
n_neighbors=n_neighbors[3:])
def test_aligned_update(aligned_iris, aligned_iris_relations):
data, target = aligned_iris
small_aligned_model = AlignedUMAP()
small_aligned_model.fit(data[:3], relations=aligned_iris_relations[:2])
small_aligned_model.update(data[3], relations=aligned_iris_relations[2])
for i, slice in enumerate(data[:4]):
data_dmat = pairwise_distances(slice)
true_nn = np.argsort(data_dmat, axis=1)[:, :10]
embd_dmat = pairwise_distances(small_aligned_model.embeddings_[i])
embd_nn = np.argsort(embd_dmat, axis=1)[:, :10]
assert nn_accuracy(true_nn, embd_nn) >= 0.65
def aligned_iris_model(aligned_iris, aligned_iris_relations):
data, target = aligned_iris
model = AlignedUMAP()
model.fit(data, relations=aligned_iris_relations)
return model
# ################################ EXTRACT EMBEDDINGS #########################
to_embed_weights = []
for layer in range(3):
stacked_w = np.vstack([
total_weights[optimizer][layer]
for optimizer in range(len(target_optimizers))
])
stacked_w = StandardScaler().fit_transform(stacked_w)
to_embed_weights.append(stacked_w)
rela = [{key: key
for key in range(to_embed_weights[0].shape[0])} for i in range(2)]
mapper = AlignedUMAP(n_components=2, metric='manhattan',
n_neighbors=30).fit(to_embed_weights, relations=rela)
emb_space_sizes = []
for emb in mapper.embeddings_:
emb_space_sizes.append([
np.append(emb.min(0),
np.array(total_train_losses).flatten().min()),
np.append(emb.max(0),
np.array(total_train_losses).flatten().max())
])
# ################################ SAVE ANIMATIONS ############################
for index, opt_name in enumerate(target_optimizers.keys()):