Machine Learning Graph Attack
The main idea of this attack is find nodes (edges) that represent same entity(person), but in two social graphs.
The algorithm is trained and tested on new orleans facebook social graph {http://socialnetworks.mpi-sws.org/data-wosn2009.html}
Basic Usage
-
Run ML-Deanon.ipynb for a model described in original paper by Kumar Sharad 2016
-
Run Edge_attack.ipynb for an extention model for attacking connections in a social graph
You can exectute it online, without installing jupyter :
Credits
The code was written using the model described in http://ksharad.com/publication/benchmarking_aisec/