Ephemeral Learning -- Augmenting Triggers with Online-Trained Normalizing Flows

TL;DR

This paper introduces a novel online data compression method for large-scale collider experiments using normalizing flows, enabling efficient offline analysis and anomaly detection of entire datasets.

Contribution

It proposes a new approach to compress entire datasets at once with normalizing flows, enhancing online trigger systems for large data rates at the LHC.

Findings

Successful demonstration on toy models

Effective anomaly detection capability

Potential for real-time data compression

Abstract

The large data rates at the LHC require an online trigger system to select relevant collisions. Rather than compressing individual events, we propose to compress an entire data set at once. We use a normalizing flow as a deep generative model to learn the probability density of the data online. The events are then represented by the generative neural network and can be inspected offline for anomalies or used for other analysis purposes. We demonstrate our new approach for a toy model and a correlation-enhanced bump hunt.