# Anomaly Detection in Grid Compute Nodes: A Machine Learning Approach Leveraging HEP Benchmark Suite

- Institution: CERN
- Department: CERN openlab Summer Student Programme 2024
- Speaker: Kacper Kamil Kozik
- Year: 2024
- Language: English
- Subjects: Computer science
- Source: [CERN](https://videos.cern.ch/record/3012160)
- Licence: [CERN](http://copyright.web.cern.ch)
- Who can watch: Tweens profiles and older (sources give no ratings; placed by Invisible Network from the source's own words)

Detecting anomalies in the performance of WLCG compute nodes is challenging and has been addressed in various ways over the years. Recently, the HEPiX Benchmarking Group introduced a novel approach using the HEP Benchmark Suite to validate WLCG compute node performance in relation to fabric metrics. By running HEPSuite on the grid, a large dataset of performance metrics has been collected. Given the volume of metrics, advanced machine learning techniques can enhance analysis. The project aims to assess how these features impact job slot performance. Project delve into benchmarking computing resources, utilization of HEPscore23, and developing a machine learning models for validating metrics, parameter correlation analysis and anomaly detection..

0:00:00 Slide 1

0:02:11 Slide 2

0:03:12 Slide 3

0:04:23 Slide 4

0:06:32 Slide 5

## Lectures

1. [Anomaly Detection in Grid Compute Nodes: A Machine Learning Approach Leveraging HEP Benchmark Suite](https://invisiblenetwork.ai/learn/cern/3012160/6bb2266a8fa0), 8 min (CERN's player)