GithubHelp home page GithubHelp logo

benedekrozemberczki / diff2vec Goto Github PK

View Code? Open in Web Editor NEW
122.0 6.0 20.0 4 MB

Reference implementation of Diffusion2Vec (Complenet 2018) built on Gensim and NetworkX.

Home Page: https://karateclub.readthedocs.io

License: GNU General Public License v3.0

Python 100.00%
diff2vec deepwalk node2vec struc2vec gensim tensorflow unsupervised-learning graph-embedding node-embedding network-embedding

diff2vec's Introduction

Diff2Vec

Arxiv codebeat badge repo size benedekrozemberczki

A graph embedding is a representation of graph vertices in a low-dimensional space, which approximately preserves properties such as distances between nodes. Vertex sequence-based embedding procedures use features extracted from linear sequences of nodes to create embeddings using a neural network. In this paper, we propose diffusion graphs as a method to rapidly generate vertex sequences for network embedding. Its computational efficiency is superior to previous methods due to simpler sequence generation, and it produces more accurate results. In experiments, we found that the performance relative to other methods improves with increasing edge density in the graph. In a community detection task, clustering nodes in the embedding space produces better results compared to other sequence-based embedding methods.

The model is now also available in the package Karate Club.

This repository provides a reference implementation for Diff2Vec as described in the paper:

Fast Sequence Based Embedding with Diffusion Graphs Benedek Rozemberczki and Rik Sarkar. International Conference on Complex Networks, 2018.


Citing

If you find Diff2Vec useful in your research, please consider citing the following paper:

>@inproceedings{rozemberczki2018fastsequence,  
  title={{Fast Sequence Based Embedding with Diffusion Graphs}},  
  author={Rozemberczki, Benedek and Sarkar, Rik},  
  booktitle={International Conference on Complex Networks},  
  year={2018},  
  pages={99--107}
 }

Requirements

The codebase is implemented in Python 3.5.2 | Anaconda 4.2.0 (64-bit).

tqdm              4.28.1
numpy             1.15.4
pandas            0.23.4
texttable         1.5.0
gensim            3.6.0
networkx          2.4
joblib            0.13.0
logging           0.4.9.6  

Datasets

The code takes an input graph in a csv file. Every row indicates an edge between two nodes separated by a comma. The first row is a header. A sample graph for the `Facebook Restaurants` dataset is included in the `data/` directory.

Options

Learning of the embedding is handled by the `src/diffusion_2_vec.py` script which provides the following command line arguments.

Input and output options

  --input    STR     Path to the edge list csv.            Default is `data/restaurant_edges.csv`
  --output   STR     Path to the embedding features.       Default is `emb/restaurant.csv`

Model options

  --model                    STR      Embedding procedure.                      Default is `non-pooled`
  --dimensions               INT      Number of embedding dimensions.           Default is 128.
  --vertex-set-cardinality   INT      Number of nodes per diffusion tree.       Default is 80.
  --num-diffusions           INT      Number of diffusions per source node.     Default is 10.
  --window-size              INT      Context size for optimization.            Default is 10.
  --iter                     INT      Number of ASGD iterations.                Default is 1.
  --workers                  INT      Number of cores.                          Default is 4.
  --alpha                    FLOAT    Initial learning rate.                    Default is 0.025.

Examples

The following commands learns a graph embedding and writes it to disk. The first column in the embedding file is the node ID.

Creating an embedding of the default dataset with the default hyperparameter settings.
python src/diffusion_2_vec.py

Creating an embedding of an other dataset the Facebook Politicians.

python src/diffusion_2_vec.py --input data/politician_edges.csv --output output/politician.csv

Creating an embedding of the default dataset in 32 dimensions, 5 sequences per source node with maximal vertex set cardinality of 40.

python src/diffusion_2_vec.py --dimensions 32 --num-diffusions 5 --vertex-set-cardinality 40

License


diff2vec's People

Contributors

benedekrozemberczki avatar mbrukman avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar  avatar

diff2vec's Issues

KeyError: "word '3' not in vocabulary"

I construct my custom edgelist.csv as instructed, and this error shows up:
2020-10-25 15:10:28,162 : INFO : training on a 8376460 raw words (8376460 effect
ive words) took 143.0s, 58587 effective words/s
Traceback (most recent call last):
File ".../src/diffusion_2_vec.py", line 121, i
n
main(args)
File ".../src/diffusion_2_vec.py", line 117, i
n main
learn_pooled_embeddings(walks, counts, args)
File ".../src/diffusion_2_vec.py", line 63, in
learn_pooled_embeddings
save_embedding(args, model, counts)
File ".../src/diffusion_2_vec.py", line 95, in
save_embedding
out.append([int(node)-1] + list(model.wv[str(node-1)]))
File "D:\Anaconda3\envs\py36\lib\site-packages\gensim\models\keyedvectors.py",
line 353, in getitem
return self.get_vector(entities)
File "D:\Anaconda3\envs\py36\lib\site-packages\gensim\models\keyedvectors.py",
line 471, in get_vector
return self.word_vec(word)
File "D:\Anaconda3\envs\py36\lib\site-packages\gensim\models\keyedvectors.py",
line 468, in word_vec
raise KeyError("word '%s' not in vocabulary" % word)
KeyError: "word '3' not in vocabulary"

How could I solve this problem?

joblib.my_exceptions.JoblibTypeError: JoblibTypeError

mldl@ub1604:~/ub16_prj/diff2vec$ python src/diffusion_2_vec.py
lapack_opt_info:
libraries = ['openblas']
library_dirs = ['/usr/local/lib']
language = f77
blas_opt_info:
libraries = ['openblas']
library_dirs = ['/usr/local/lib']
language = f77
openblas_info:
libraries = ['openblas']
library_dirs = ['/usr/local/lib']
language = f77
openblas_lapack_info:
libraries = ['openblas']
library_dirs = ['/usr/local/lib']
language = f77
blas_mkl_info:
NOT AVAILABLE


Feature extraction starts.

0%| | 0/10 [00:00<?, ?it/s]Traceback (most recent call last):
File "src/diffusion_2_vec.py", line 60, in
main(args)
File "src/diffusion_2_vec.py", line 54, in main
walks = run_parallel_feature_creation(args.input, args.vertex_set_cardinality, args.num_diffusions, args.workers)
File "src/diffusion_2_vec.py", line 35, in run_parallel_feature_creation
results = Parallel(n_jobs = num_of_workers)(delayed(create_features)(i, edge_list_path, vertex_set_cardinality) for i in tqdm(range(number_of_replicates)))
File "/usr/local/lib/python2.7/dist-packages/joblib/parallel.py", line 789, in call
self.retrieve()
File "/usr/local/lib/python2.7/dist-packages/joblib/parallel.py", line 740, in retrieve
raise exception
joblib.my_exceptions.JoblibTypeError: JoblibTypeError

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.