lle.py 文件源码

def learn_embedding(self, graph=None, edge_f=None,
                        is_weighted=False, no_python=False):
        if not graph and not edge_f:
            raise Exception('graph/edge_f needed')
        if not graph:
            graph = graph_util.loadGraphFromEdgeListTxt(edge_f)
        graph = graph.to_undirected()
        t1 = time()
        A = nx.to_scipy_sparse_matrix(graph)
        normalize(A, norm='l1', axis=1, copy=False)
        I_n = sp.eye(graph.number_of_nodes())
        I_min_A = I_n - 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)