This repository contains our project for the course in Automatic Verification of Intelligent Systems and the AFC of Sapienza University.
The objective is the implementation of a system capable of online detection of anomalies in high dimensional data.
Depending on the intrinsic dimensionality of the data, the algorithm chooses either a GEM based or a PCA based strategy to extract summary statistics and decide if an anomaly was detected.
The work is based on the paper Real-Time Nonparametric Anomaly Detection in High-Dimensional Settings
For a detailed report read the full report.