A5003381899- Particle physics theoretical and experimental studies
- High-Energy Particle Collisions Research
- Quantum Chromodynamics and Particle Interactions
- Particle Detector Development and Performance
- Dark Matter and Cosmic Phenomena
- Computational Physics and Python Applications
- Cosmology and Gravitation Theories
- Neutrino Physics Research
- Scientific Computing and Data Management
- Astrophysics and Cosmic Phenomena
- Distributed and Parallel Computing Systems
- Atomic and Subatomic Physics Research
- Research Data Management Practices
- Black Holes and Theoretical Physics
- Adversarial Robustness in Machine Learning
- Explainable Artificial Intelligence (XAI)
- Parallel Computing and Optimization Techniques
- Big Data Technologies and Applications
- Nuclear reactor physics and engineering
- Radiation Detection and Scintillator Technologies
- Noncommutative and Quantum Gravity Theories
- Neural Networks and Applications
- Optical properties and cooling technologies in crystalline materials
- Generative Adversarial Networks and Image Synthesis
- Anomaly Detection Techniques and Applications
University of California, San Diego
2021-2025
University of California System
2024-2025
Institute of High Energy Physics
2022-2024
A. Alikhanyan National Laboratory
2022-2024
University of Antwerp
2024
Universidad Católica Santo Domingo
2023
To enable the reusability of massive scientific datasets by humans and machines, researchers aim to adhere principles findability, accessibility, interoperability, (FAIR) for data artificial intelligence (AI) models. This article provides a domain-agnostic, step-by-step assessment guide evaluate whether or not given dataset meets these principles. We demonstrate how use this FAIRness an open simulated produced CMS Collaboration at CERN Large Hadron Collider. consists Higgs boson decays quark...
Abstract We provide details on the implementation of a machine-learning based particle flow algorithm for CMS. The standard reconstructs stable particles calorimeter clusters and tracks to global event reconstruction that exploits combined information multiple detector subsystems, leading strong improvements quantities such as jets missing transverse energy. have studied possible evolution towards heterogeneous computing platforms GPUs using graph neural network. machine-learned PF model...
Abstract Efficient and accurate algorithms are necessary to reconstruct particles in the highly granular detectors anticipated at High-Luminosity Large Hadron Collider Future Circular Collider. We study scalable machine learning models for event reconstruction electron-positron collisions based on a full detector simulation. Particle-flow can be formulated as supervised task using tracks calorimeter clusters. compare graph neural network kernel-based transformer demonstrate that we avoid...
Abstract The findable, accessible, interoperable, and reusable (FAIR) data principles provide a framework for examining, evaluating, improving how is shared to facilitate scientific discovery. Generalizing these research software other digital products an active area of research. Machine learning models—algorithms that have been trained on without being explicitly programmed—and more generally, artificial intelligence (AI) models, are important target this because the ever-increasing pace...
Autoencoders have useful applications in high energy physics anomaly detection, particularly for jets - collimated showers of particles produced collisions such as those at the CERN Large Hadron Collider. We explore use graph-based autoencoders, which operate on their "particle cloud" representations and can leverage interdependencies among within a jet, tasks. Additionally, we develop differentiable approximation to mover's distance via graph neural network, may subsequently be used...
The particle-flow (PF) algorithm is used in general-purpose particle detectors to reconstruct a comprehensive particle-level view of the collision by combining information from different subdetectors. A graph neural network (GNN) model, known as machine-learned (MLPF) algorithm, has been developed substitute rule-based PF algorithm. However, understanding model's decision making not straightforward, especially given complexity set-to-set prediction task, dynamic building, and message-passing...
The particle-flow (PF) algorithm, which infers particles based on tracks and calorimeter clusters, is of central importance to event reconstruction in the CMS experiment at CERN LHC, has been a focus development light planned Phase-2 running conditions with an increased pileup detector granularity. In recent years, machine learned (MLPF) graph neural network that performs PF reconstruction, explored CMS, possible advantages directly optimizing for physical quantities interest, being highly...
At the CERN LHC, task of jet tagging, whose goal is to infer origin a given set final-state particles, dominated by machine learning methods. Graph neural networks have been used address this treating jets as point clouds with underlying, learnable, edge connections between particles inside. We explore decision-making process for one such state-of-the-art network, ParticleNet, looking relevant identified using layerwise-relevance propagation technique. As model trained, we observe changes in...
The findable, accessible, interoperable, and reusable (FAIR) data principles serve as a framework for examining, evaluating, improving sharing to advance scientific endeavors. There is an emerging trend adapt these machine learning models—algorithms that learn from without specific coding—and, more generally, AI models, due AI’s swiftly growing impact on engineering sectors. In this paper, we propose practical definition of the FAIR models provide template program their adoption. We...
In high energy physics, self-supervised learning (SSL) methods have the potential to aid in creation of machine models without need for labeled datasets a variety tasks, including those related jets -- narrow sprays particles produced by quarks and gluons particle collisions. This study introduces an approach jet representations hand-crafted augmentations using jet-based joint embedding predictive architecture (J-JEPA), which aims predict various physical targets from informative context. As...
Abstract We study scalable machine learning models for full event reconstruction in high-energy electron-positron collisions based on a highly granular detector simulation. Particle-flow (PF) can be formulated as supervised task using tracks and calorimeter clusters or hits. compare graph neural network kernel-based transformer demonstrate that both avoid quadratic memory allocation computational cost while achieving realistic PF reconstruction. show hyperparameter tuning supercomputer...
This study introduces an innovative approach to analyzing unlabeled data in high-energy physics (HEP) through the application of self-supervised learning (SSL). Faced with increasing computational cost producing high-quality labeled simulation samples at CERN LHC, we propose leveraging large volumes overcome limitations supervised methods, which heavily rely on detailed simulations. By pretraining models these vast, mostly untapped datasets, aim learn generic representations that can be...
In this community review report, we discuss applications and techniques for fast machine learning (ML) in science -- the concept of integrating power ML methods into real-time experimental data processing loop to accelerate scientific discovery. The material report builds on two workshops held by Fast Science covers three main areas: across a number domains; training implementing performant resource-efficient algorithms; computing architectures, platforms, technologies deploying these...
We provide details on the implementation of a machine-learning based particle flow algorithm for CMS. The standard reconstructs stable particles calorimeter clusters and tracks to global event reconstruction that exploits combined information multiple detector subsystems, leading strong improvements quantities such as jets missing transverse energy. have studied possible evolution towards heterogeneous computing platforms GPUs using graph neural network. machine-learned PF model candidates...
The findable, accessible, interoperable, and reusable (FAIR) data principles provide a framework for examining, evaluating, improving how is shared to facilitate scientific discovery. Generalizing these research software other digital products an active area of research. Machine learning (ML) models -- algorithms that have been trained on without being explicitly programmed more generally, artificial intelligence (AI) models, are important target this because the ever-increasing pace with...