from time import time
import networkx as nx
import numpy as np
import scipy.io as sio
import scipy.sparse as sp
import scipy.sparse.linalg as lg
from . import graph as g
from sklearn.preprocessing import normalize
__author__ = "Alan WANG"
__email__ = "alan1995wang@outlook.com"
class LLE(object):
def __init__(self, graph, d):
''' Initialize the LocallyLinearEmbedding class
Args:
graph: nx.DiGraph
input Graph
d: int
dimension of the embedding
'''
self._d = d
self._method_name = "lle_svd"
self.g = graph
self._node_num = graph.node_size
self.learn_embedding()
def learn_embedding(self):
graph = self.g.G
graph = graph.to_undirected()
t1 = time()
A = nx.to_scipy_sparse_matrix(graph)
# print(np.sum(A.todense(), axis=0))
normalize(A, norm='l1', axis=1, copy=False)
I_n = sp.eye(graph.number_of_nodes())
I_min_A = I_n - A
print(I_min_A)
u, s, vt = lg.svds(I_min_A, k=self._d + 1, which='SM')
t2 = time()
self._X = vt.T
self._X = self._X[:, 1:]
return self._X, (t2 - t1)
# I_n = sp.eye(graph.number_of_nodes())
@property
def vectors(self):
vectors = {}
look_back = self.g.look_back_list
for i, embedding in enumerate(self._X):
vectors[look_back[i]] = embedding
return vectors
def save_embeddings(self, filename):
fout = open(filename, 'w')
node_num = len(self.vectors.keys())
fout.write("{} {}\n".format(node_num, self._d))
for node, vec in self.vectors.items():
fout.write("{} {}\n".format(node,
' '.join([str(x) for x in vec])))
fout.close()
