abojchevski / paican

Implementation of "Bayesian Robust Attributed Graph Clustering: Joint Learning of Partial Anomalies and Group Structure".

Paican

Tensorflow implementation of the method proposed in the paper: "Bayesian Robust Attributed Graph Clustering: Joint Learning of Partial Anomalies and Group Structure", Aleksandar Bojchevski and Stephan Günnemann, AAAI 2018.

Installation

python setup.py install

Requirements

• tensorflow (>=1.4, <=2.0)
• sklearn (only for evaluation)

Note: If you are using tensorflow >=2.0 you can stull run the above code by replacing the tensorflow import with

import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()

and changing tf.contrib.distributions to tf.distributions.

Data

Each of the dataset folders consists of the following files:

• A.mtx - the adjacency matrix in scipy's sparse csr_matrix format
• X.mtx - the attribute matrix in scipy's sparse csr_matrix format
• feature_to_index.npy - a dictionary mapping a feature label to index (e.g. 'neurology' -> 5)
• node_to_index.npy - a dictionary mapping a node label (e.g name of a person or paper ID) to index
• z.npy - ground truth clusters if available
• label_to_cluster.npy - a dictionary mapping a label (e.g. journal, party) to cluster index

Demo

• See the notebook example.ipynb for a simple demo.
• Visit our website for an interactive plot that shows the inferred clustering on a subset of the Amazon dataset.

Cite

Please cite our paper if you use this code in your own work.