Synthetic Data Generation with Lorenzetti for Time Series Anomaly Detection in High-Energy Physics Calorimeters

2025-09-09

AI Summary

Anomaly detection in high-energy physics calorimeters faces challenges due to scarce labeled data, unknown anomaly types, and strong cross-dimensional correlations in multivariate time series. Method: We propose a physics-informed synthetic data augmentation framework leveraging the Lorenzetti simulator to flexibly inject diverse defect patterns, generating high-fidelity synthetic datasets covering varied detector configurations and anomaly classes. Building on this, we design a time-series-optimized Transformer architecture and systematically evaluate its sensitivity to complex, cross-dimensional anomaly patterns. Results: Experiments demonstrate significant improvements in model generalization and robustness under low-labeling regimes. Our approach enables interpretable and reproducible anomaly localization, supporting practical deployment. To our knowledge, this is the first work to deeply integrate high-fidelity physics simulation into the end-to-end pipeline of calorimeter time-series anomaly detection.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Anomaly detection in multivariate time series is crucial to ensure the quality of data coming from a physics experiment. Accurately identifying the moments when unexpected errors or defects occur is essential, yet challenging due to scarce labels, unknown anomaly types, and complex correlations across dimensions. To address the scarcity and unreliability of labelled data, we use the Lorenzetti Simulator to generate synthetic events with injected calorimeter anomalies. We then assess the sensitivity of several time series anomaly detection methods, including transformer-based and other deep learning models. The approach employed here is generic and applicable to different detector designs and defects.

Problem

• Detecting anomalies in multivariate time series data
• Addressing scarce and unreliable labeled anomaly data
• Evaluating anomaly detection methods for calorimeter defects

Innovation

• Uses Lorenzetti Simulator for synthetic anomaly generation
• Evaluates transformer-based and deep learning models
• Generic approach applicable to various detector designs