allentran / graph2vec

Theano implementation of GloVe for graphs

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

graph2vec

Learning node embeddings via Theano (the GPU if you have a good one), with minibatch and AdaGrad.

Install via running

python setup.py install

or

pip install graph2vec

Data should be in space delimited files describing edges, either as from_node to_node distance or without the path distance. For example, in a text file like

0 1
1 3
14 21
21 1

Use the helper trainer module to build/load the graph.

import graph2vec.trainer

graph2vec = Graph2Vec(vector_dimensions=128)
graph2vec.parse_graph('edge.data', extend_paths=2)

The extend_paths argument dictates the length of paths included in the cost function. Note that only the shortest path between nodes are included. Fit the vectors with

graph2vec.fit(batch_size=1000, max_epochs=1000)

Don't go too crazy with the batch size, you'll get a speed up but convergence will be erratic. The vectors, both as origin and destination nodes are stored in the model object.

graph2vec.model.Win.get_value()
graph2vec.model.Wout.get_value()

About

Theano implementation of GloVe for graphs

License:Apache License 2.0


Languages

Language:Python 100.0%