Feature Learning and Network Structure from Noisy Node Activity Data

TL;DR

This paper introduces an unsupervised learning framework that constructs networks and learns node representations from noisy, incomplete node activity data, enabling network analysis without explicit edge information.

Contribution

It proposes a novel method combining random node sequence generation, neural network training, and entropy-based neighbor selection to analyze noisy node activity data.

Findings

Effective in synthetic data scenarios

Validated on real-world datasets

Captures nodes with synergistic roles

Abstract

In the studies of network structures, much attention has been devoted to developing approaches to reconstruct networks and predict missing links when edge-related information is given. However, such approaches are not applicable when we are only given noisy node activity data with missing values. This work presents an unsupervised learning framework to learn node vectors and construct networks from such node activity data. First, we design a scheme to generate random node sequences from node context sets, which are generated from node activity data. Then, a three-layer neural network is adopted training the node sequences to obtain node vectors, which allow us to construct networks and capture nodes with synergistic roles. Furthermore, we present an entropy-based approach to select the most meaningful neighbors for each node in the resulting network. Finally, the effectiveness of the method is validated through both synthetic and real data.