Hide and Seek: Outwitting Community Detection Algorithms

Community affiliation of a node plays an important role in determining its contextual position the network, which may raise privacy concerns when sensitive wants to hide identity network. Oftentimes, target community seeks protect itself from adversaries so that constituent members remain hidden inside The current study focuses on hiding such communities targeted nodes can be concealed. This leads problem deception, investigates avenues minimally rewiring network given maximally hides detection algorithm (CDA). We formalize deception and introduce Network using permanence loss (NEURAL), novel method greedily optimizes node-centric objective function determine strategy. Theoretical settings pose restriction number strategies employed optimize function, turn reduces overhead choosing best strategy multiple options. also show our is submodular monotone. When tested both synthetic seven real-world networks, NEURAL able deceive six widely used CDAs. benchmark performance with respect four state-of-the-art methods evaluation metrics. In addition, qualitative analysis three other attributed networks reveals NEURAL, quite strikingly, captures metainformation about edges otherwise could not inferred by observing only their topological structures.