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.