Capability benchmark · Physics · Computer science
LHC Olympics benchmarks anomaly detection for collider physics
Community benchmark for model-agnostic anomaly detection in simulated collider events.
Summary
The arXiv paper reviews the LHC Olympics 2020 community challenge for model-agnostic searches for new physics. It describes simulated collider-event datasets, hidden black-box test sets, the anomaly-detection methods participants used and lessons for future collider analyses.
AI role
Compared machine-learning approaches for finding unexpected signals in high-energy physics data.
Narrative role
LHC Olympics is supporting evidence for how AI changes analysis strategy in data-intensive physics.
Caveat
Benchmark success on simulated events does not guarantee discovery of new physics in real collider data.