15–18 Oct 2024
Purdue University
America/Indiana/Indianapolis timezone

Autonomous discoveries using a modular ecosystem for adaptive anomaly detection in LHC triggers

Not scheduled
20m
Steward Center 306 (Third floor) (Purdue University)

Steward Center 306 (Third floor)

Purdue University

128 Memorial Mall Dr, West Lafayette, IN 47907
Poster

Speaker

Shaghayegh Emami

Description

Anomaly detection (AD) in the earliest stage of LHC trigger systems represents a fundamentally new tool to enable data-driven discoveries. While initial efforts have focused on adapting powerful offline algorithms to these high-throughput streaming systems, the question of how such algorithms should adapt to constantly-evolving detector conditions remains a major challenge. In this work, we introduce a modular ecosystem to develop and assess strategies for autonomous discovery that incorporates diverse components including: datasets with time-dependent effects, complex trigger menus, real-time control mechanisms, and cost-aware optimization criteria. We illustrate this framework with a novel benchmark based on reinforcement learning for AD triggers using public CMS datasets, aiming to encourage community-driven development towards a new generation of both intelligent and adaptive triggers.

Focus areas HEP

Authors

Abhijith Gandrakota (Fermi National Accelerator Lab.(US)) Cecilia Tosciri (University of Chicago (US)) Christian Herwig (University of Michigan (US)) David Miller (University of Chicago (US)) Jennifer Ngadiuba (FNAL) Nhan Tran (Fermi National Accelerator Lab. (US)) Shaghayegh Emami Prof. Yuxin Chen (Univ. of Chicago) Zixin Ding (Univ. of Chicago)

Presentation materials