Unsupervised robust nonparametric learning of hidden community properties

TL;DR

This paper introduces a scalable, nonparametric, and unsupervised graph scan method for detecting community properties in large noisy networks, robust against adversarial manipulation, with proven consistency and real-world validation.

Contribution

It presents a novel robust graph scan procedure for community detection that is unsupervised, scalable, and resistant to adversaries, with theoretical guarantees.

Findings

Proven strong consistency of the method under minimal assumptions.

Effective detection of community properties even with adversarial influence.

Validated performance on real and synthetic datasets.

Abstract

We consider learning of fundamental properties of communities in large noisy networks, in the prototypical situation where the nodes or users are split into two classes according to a binary property, e.g., according to their opinions or preferences on a topic. For learning these properties, we propose a nonparametric, unsupervised, and scalable graph scan procedure that is, in addition, robust against a class of powerful adversaries. In our setup, one of the communities can fall under the influence of a knowledgeable adversarial leader, who knows the full network structure, has unlimited computational resources and can completely foresee our planned actions on the network. We prove strong consistency of our results in this setup with minimal assumptions. In particular, the learning procedure estimates the baseline activity of normal users asymptotically correctly with probability 1; the only assumption being the existence of a single implicit community of asymptotically negligible logarithmic size. We provide experiments on real and synthetic data to illustrate the performance of our method, including examples with adversaries.