A5004908171- Particle physics theoretical and experimental studies
- Particle Detector Development and Performance
- High-Energy Particle Collisions Research
- Computational Physics and Python Applications
- Machine Learning and Data Classification
- Advanced Neural Network Applications
- Machine Learning and Algorithms
- Astrophysics and Cosmic Phenomena
- Advanced Multi-Objective Optimization Algorithms
- Neutrino Physics Research
- Medical Imaging Techniques and Applications
University of California, Irvine
2014-2022
Classification of jets as originating from light-flavor or heavy-flavor quarks is an important task for inferring the nature particles produced in high-energy collisions. The large and variable dimensionality data provided by tracking detectors makes this difficult. current state-of-the-art tools require expert reduction to convert into a fixed low-dimensional form that can be effectively managed shallow classifiers. We study application deep networks task, attempting classification at...
Sherpa is a hyperparameter optimization library for machine learning models. It specifically designed problems with computationally expensive, iterative function evaluations, such as the tuning of deep neural networks. With Sherpa, scientists can quickly optimize hyperparameters using variety powerful and interchangeable algorithms. be run on either single or in parallel cluster. Finally, an interactive dashboard enables users to view progress models they are trained, cancel trials, explore...
QCD-jets at the LHC are described by simple physics principles. We show how super-resolution generative networks can learn underlying structures and use them to improve resolution of jet images. test this approach on massless fat top-jets find that network reproduces their main features even without training pure samples. In addition, we a slim architecture be constructed once have control full performance.
Generation of simulated data is essential for analysis in particle physics, but current Monte Carlo methods are very computationally expensive. Deep-learning-based generative models have successfully generated at lower cost, struggle when the sparse. We introduce a novel deep sparse autoregressive model (SARM) that explicitly learns sparseness with tractable likelihood, making it more stable and interpretable compared to Generative Adversarial Networks (GANs) other methods. In two case...
We investigate whether state-of-the-art classification features commonly used to distinguish electrons from jet backgrounds in collider experiments are overlooking valuable information. A deep convolutional neural network analysis of electromagnetic and hadronic calorimeter deposits is compared the performance typical features, revealing a $\approx 5\%$ gap which indicates that these lower-level data do contain untapped power. To reveal nature this unused information, we use recently...
The classification of jets as quark- versus gluon-initiated is an important yet challenging task in the analysis data from high-energy particle collisions and search for physics beyond Standard Model. recent integration deep neural networks operating on low-level detector information has resulted significant improvements power quark/gluon jet tagging models. However, improved such models trained simulated samples come at cost reduced interpretability, raising concerns about their...
QCD-jets at the LHC are described by simple physics principles. We show how super-resolution generative networks can learn underlying structures and use them to improve resolution of jet images. test this approach on massless fat top-jets find that network reproduces their main features even without training pure samples. In addition, we a slim architecture be constructed once have control full performance.
Distinguishing between prompt muons produced in heavy boson decay and association with heavy-flavor jet production is an important task analysis of collider physics data. We explore whether there information available calorimeter deposits that not captured by the standard approach isolation cones. find convolutional networks particle-flow accessing cells surpass performance cones, suggesting radial energy distribution angular structure surrounding muon contain unused discrimination power....
In the framework of three-active-neutrino mixing, charge parity phase, neutrino mass ordering, and octant $\theta_{23}$ remain unknown. The Deep Underground Neutrino Experiment (DUNE) is a next-generation long-baseline oscillation experiment, which aims to address these questions by measuring patterns $\nu_\mu/\nu_e$ $\bar\nu_\mu/\bar\nu_e$ over range energies spanning first second maxima. DUNE far detector modules are based on liquid argon TPC (LArTPC) technology. A LArTPC offers excellent...
Sherpa is a hyperparameter optimization library for machine learning models. It specifically designed problems with computationally expensive, iterative function evaluations, such as the tuning of deep neural networks. With Sherpa, scientists can quickly optimize hyperparameters using variety powerful and interchangeable algorithms. be run on either single or in parallel cluster. Finally, an interactive dashboard enables users to view progress models they are trained, cancel trials, explore...