A5107979834- Particle physics theoretical and experimental studies
- High-Energy Particle Collisions Research
- Particle Detector Development and Performance
- Quantum Chromodynamics and Particle Interactions
- Computational Physics and Python Applications
- Dark Matter and Cosmic Phenomena
- Cosmology and Gravitation Theories
- Neutrino Physics Research
- Distributed and Parallel Computing Systems
- Astrophysics and Cosmic Phenomena
- Medical Imaging Techniques and Applications
- Algorithms and Data Compression
- Radiation Detection and Scintillator Technologies
- Black Holes and Theoretical Physics
- Pulsars and Gravitational Waves Research
- Advanced Data Storage Technologies
- advanced mathematical theories
- Superconducting Materials and Applications
- Anomaly Detection Techniques and Applications
- Scientific Computing and Data Management
- Gaussian Processes and Bayesian Inference
- Machine Learning in Materials Science
- Machine Learning and Algorithms
- Gamma-ray bursts and supernovae
- Adversarial Robustness in Machine Learning
University of California, Irvine
2016-2025
University of California System
2025
University of California, Santa Cruz
2018-2024
University of Geneva
2024
Istanbul University
2024
SR Research (Canada)
2024
Federación Española de Enfermedades Raras
2024
A. Alikhanyan National Laboratory
2024
The University of Adelaide
2016-2023
Rutherford Appleton Laboratory
2011-2023
We investigate a new structure for machine learning classifiers built with neural networks and applied to problems in high-energy physics by expanding the inputs include not only measured features but also parameters. The parameters represent smoothly varying task, resulting parameterized classifier can interpolate between them replace sets of trained at individual values. This simplifies training process gives improved performance intermediate values, even complex requiring deep learning....
At the extreme energies of Large Hadron Collider, massive particles can be produced at such high velocities that their hadronic decays are collimated and resulting jets overlap. Deducing whether substructure an observed jet is due to a low-mass single particle or multiple decay objects important problem in analysis collider data. Traditional approaches have relied on expert features designed detect energy deposition patterns calorimeter, but complexity data make this task excellent candidate...
Machine learning has played an important role in the analysis of high-energy physics data for decades. The emergence deep 2012 allowed machine tools which could adeptly handle higher-dimensional and more complex problems than previously feasible. This review is aimed at reader who familiar with high energy but not learning. connections between are explored, followed by introduction to core concepts neural networks, examples key results demonstrating power LHC data, discussion future...
First-principle simulations are at the heart of high-energy physics research program. They link vast data output multi-purpose detectors with fundamental theory predictions and interpretation. This review illustrates a wide range applications modern machine learning to event generation simulation-based inference, including conceptional developments driven by specific requirements particle physics. New ideas tools developed interface will improve speed precision forward simulations, handle...
Recent innovations from machine learning allow for data unfolding, without binning and including correlations across many dimensions. We describe a set of known, upgraded, new methods ML-based unfolding. The performance these approaches are evaluated on the same two datasets. find that all techniques capable accurately reproducing particle-level spectra complex observables. Given conceptually diverse, they offer an exciting toolkit class measurements can probe Standard Model with...
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...
We describe a strategy for constructing neural network jet substructure tagger which powerfully discriminates boosted decay signals while remaining largely uncorrelated with the mass. This reduces impact of systematic uncertainties in background modeling enhancing signal purity, resulting improved discovery significance relative to existing taggers. The is trained using an adversarial strategy, that learns balance classification accuracy decorrelation. As benchmark scenario, we consider case...
Searches for dark matter at colliders typically involve signatures with energetic initial-state radiation without visible recoil particles. mono-jet or mono-photon have yielded powerful constraints on interactions Standard Model We extend this to the mono-Z signature and reinterpret an ATLAS analysis of events a Z boson missing transverse momentum derive interaction mass scale nucleon cross sections in context effective field theories describing which interacts via heavy mediator particles...
We explore the LHC phenomenology of dark matter (DM) pair production in association with a 125 GeV Higgs boson. This signature, dubbed `mono-Higgs,' appears as single boson plus missing energy from DM particles escaping detector. perform an background study for mono-Higgs signals at $\sqrt{s} = 8$ and $14$ TeV four decay channels: $\gamma\gamma$, $b \bar b$, $ZZ^* \to 4\ell$, $\ell\ell j j$. estimate sensitivities to variety new physics scenarios within frameworks both effective operators...
The Higgs boson is thought to provide the interaction that imparts mass fundamental fermions, but while measurements at Large Hadron Collider (LHC) are consistent with this hypothesis, current analysis techniques lack statistical power cross traditional $5\ensuremath{\sigma}$ significance barrier without more data. Deep learning have potential increase of by automatically complex, high-level data representations. In work, deep neural networks used detect decay a pair tau leptons. A Bayesian...
Abstract Neutron stars provide a unique laboratory for studying matter at extreme pressures and densities. While there is no direct way to explore their interior structure, X-rays emitted from these can indirectly clues the equation of state (EOS) superdense nuclear through inference star's mass radius. However, EOS directly X-ray spectra extremely challenging complicated by systematic uncertainties. The current art use simulation-based likelihoods in piece-wise method which relies on...
Lines in the spectrum of cosmic gamma rays are considered one more robust signatures dark matter annihilation. We consider such processes from an effective field theory vantage, and find that generically, two or lines expected, providing interesting feature can be exploited for searches reveal details about underlying matter. Using 130 GeV recently reported Fermi-LAT data as example, we analyze energy multi-line context to consistent with a single γγ line, γZ line both line.
We present a technique for translating black-box machine-learned classifier operating on high-dimensional input space into small set of human-interpretable observables that can be combined to make the same classification decisions. iteratively select these from large high-level discriminants by finding those with highest decision similarity relative black box, quantified via metric we introduce evaluates ordering pairs inputs. Successive iterations focus only subset are misordered current...
Machine learning techniques are becoming an integral component of data analysis in high energy physics. These tools provide a significant improvement sensitivity over traditional analyses by exploiting subtle patterns high-dimensional feature spaces. may not be well modeled the simulations used for training machine methods, resulting enhanced to systematic uncertainties. Contrary wisdom constructing strategy that is invariant uncertainties, we study use classifier fully aware uncertainties...
Top quarks, produced in large numbers at the Large Hadron Collider, have a complex detector signature and require special reconstruction techniques. The most common decay mode, ``all-jet'' channel, results 6-jet final-state which is particularly difficult to reconstruct $pp$ collisions due number of permutations possible. We present novel approach this class problem, based on neural networks using generalized attention mechanism, that we call symmetry preserving (spa-net). train one such...
In many scientific fields which rely on statistical inference, simulations are often used to map from theoretical models experimental data, allowing scientists test model predictions against results. Experimental data is reconstructed indirect measurements causing the aggregate transformation be poorly-described analytically. Instead, numerical at great computational cost. We introduce Optimal-Transport-based Unfolding and Simulation (OTUS), a fast simulator based unsupervised...
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.
Abstract Neutron stars provide a unique opportunity to study strongly interacting matter under extreme density conditions. The intricacies of inside neutron and their equation state are not directly visible, but determine bulk properties, such as mass radius, which affect the star's thermal X-ray emissions. However, telescope spectra these emissions also affected by stellar distance, hydrogen column, effective surface temperature, always well-constrained. Uncertainties on nuisance parameters...
The ATLAS Phase-I upgrade (2019) requires a Trigger and Data Acquisition (TDAQ) system able to trigger record data from up three times the nominal LHC instantaneous luminosity. Front-End LInk eXchange (FELIX) provides an infrastructure achieve this in scalable, detector agnostic easily upgradeable way. It is PC-based gateway, interfacing custom radiation tolerant optical links front-end electronics, via PCIe Gen3 cards, commodity switched Ethernet or InfiniBand network. FELIX enables...
High-energy collisions at the Large Hadron Collider (LHC) provide valuable insights into open questions in particle physics. However, detector effects must be corrected before measurements can compared to certain theoretical predictions or from other detectors. Methods solve this \textit{inverse problem} of mapping observations quantities underlying collision are essential parts many physics analyses LHC. We investigate and compare various generative deep learning methods approximate inverse...
Searches for dark matter production at particle colliders are complementary to direct-detection and indirect-detection experiments especially powerful small masses, ${m}_{\ensuremath{\chi}}<100\text{ }\text{ }\mathrm{GeV}$. An important collider signature is due the of a pair these invisible particles with initial-state radiation standard model particle. Currently, searches use individual nearly orthogonal final states search jets, photons or massive gauge bosons. We combine results across...
We analyze the potential dark matter implications of LHC events with missing transverse momentum and a resonance, such as Z', decaying to pair jets or leptons. This final state contains significant discovery potential, but has not yet been examined in detail by experiments. introduce models Z' production association particles, propose reconstruction selection strategies, estimate sensitivity current dataset.
Abstract Reconstructing unstable heavy particles requires sophisticated techniques to sift through the large number of possible permutations for assignment detector objects underlying partons. An approach based on a generalized attention mechanism, symmetry preserving networks (SPA-NET), has been previously applied top quark pair decays at Large Hadron Collider which produce only hadronic jets. Here we extend SPA-NET architecture consider multiple input object types, such as leptons, well...