Recent years have witnessed an upsurge of interest in the problem of anomalydetection on attributed networks . Hop counts based on both global and local contextual information can be served as indicators of anomaly . We propose a hop-countbased model (HCM) to detect anomalies by modeling both local and global contextual information . We design two anomaly scores based on the hop counts prediction via HCMmodel to identify anomalies . We employ Bayesian learning to train HCM model for capturing uncertainty in learned parameters and avoiding overfitting. Besides, we use Bayesian Learning to train the model to capture uncertainty in learning parameters and avoid overfitting, we propose to use hop counts predictions as a self-supervised task. We also design two anomalies based on hop counts to identify anomaly scores. Besides Bayesianlearning to train our new anomaly scores to identify an anomaly. It is effective in using our new model. It was effective in testing our new data sets. It has been effective in real-world attributed networks. It’s effective in detecting anomalies. It provides an efficient way to detect anomaly.

Author(s) : Tianjin Huang, Yulong Pei, Vlado Menkovski, Mykola Pechenizkiy

Links : PDF - Abstract

Code :
Coursera

Keywords : anomaly - hop - effective - anomalies - counts -

Leave a Reply

Your email address will not be published. Required fields are marked *