Skip to content

Anomaly Detection on real High-Performance-Computing time-series, using Autoencoder Neural Network

Notifications You must be signed in to change notification settings

RiccardoEvangelisti/HPC-Anomaly-Detection

Repository files navigation

Anomaly Detection on High-Performance Computing real data using Autoencoders

The study focuses on using Autoencoder Neural Networks for Anomaly Detection in High-Performance Computing (HPC) data from the Marconi100 supercomputer. The main objective is to develop a model capable of differentiating between normal and anomalous node states and to detect anomalies before manual checks by system administrators. The research highlights the effectiveness of an overcomplete autoencoder, which achieved high precision, recall, and F1 scores, demonstrating the model's accuracy in identifying anomalies. Notably, the model was able to detect anomalies up to 90 minutes before the manual system administrators intervention.

Usage

  • Report_HPC_Anomaly_Detection.pdf describes the resarch in detail
  • Data_Exploration.ipynb contains the analysis and resampling of the original data
  • The original data is downloadable here
  • semi_supervised/ folder contains the program
  • papers/ folder contains articles on the topic and code examples
  • query_tool is a python module, necessary for manipulating the original data

About

Anomaly Detection on real High-Performance-Computing time-series, using Autoencoder Neural Network

Topics

Resources

Stars

Watchers

Forks

Languages