More Like this

1. Search for new non-resonant phenomena in high-mass dilepton final states with the ATLAS detector

   https://doi.org/10.1007/JHEP11(2020)005  

   Aad, G.; Abbott, B.; Abbott, D. C.; Abed Abud, A.; Abeling, K.; Abhayasinghe, D. K.; Abidi, S. H.; AbouZeid, O. S.; Abraham, N. L.; Abramowicz, H.; et al (November 2020, Journal of High Energy Physics)

   null (Ed.)

   A bstract A search for new physics with non-resonant signals in dielectron and dimuon final states in the mass range above 2 TeV is presented. This is the first search for non-resonant signals in dilepton final states at the LHC to use a background estimate from the data. The data, corresponding to an integrated luminosity of 139 fb − 1 , were recorded by the ATLAS experiment in proton-proton collisions at a center-of-mass energy of $$ \sqrt{s} $$ s = 13 TeV during Run 2 of the Large Hadron Collider. The benchmark signal signature is a two-quark and two-lepton contact interaction, which would enhance the dilepton event rate at the TeV mass scale. To model the contribution from background processes a functional form is fit to the dilepton invariant-mass spectra in data in a mass region below the region of interest. It is then extrapolated to a high-mass signal region to obtain the expected background there. No significant deviation from the expected background is observed in the data. Upper limits at 95% CL on the number of events and the visible cross-section times branching fraction for processes involving new physics are provided. Observed (expected) 95% CL lower limits on the contact interaction energy scale reach 35.8 (37.6) TeV. 

   more » « less
2. Anomaly detection in the presence of irrelevant features

   https://doi.org/10.1007/JHEP02(2024)220  

   Freytsis, Marat; Perelstein, Maxim; San, Yik Chuen (February 2024, Journal of High Energy Physics)

   A<sc>bstract</sc> Experiments at particle colliders are the primary source of insight into physics at microscopic scales. Searches at these facilities often rely on optimization of analyses targeting specific models of new physics. Increasingly, however, data-driven model-agnostic approaches based on machine learning are also being explored. A major challenge is that such methods can be highly sensitive to the presence of many irrelevant features in the data. This paper presents Boosted Decision Tree (BDT)-based techniques to improve anomaly detection in the presence of many irrelevant features. First, a BDT classifier is shown to be more robust than neural networks for the Classification Without Labels approach to finding resonant excesses assuming independence of resonant and non-resonant observables. Next, a tree-based probability density estimator using copula transformations demonstrates significant stability and improved performance over normalizing flows as irrelevant features are added. The results make a compelling case for further development of tree-based algorithms for more robust resonant anomaly detection in high energy physics. 

   more » « less
3. An Oceanic Ultra-Violet Catastrophe, Wave-Particle Duality and a Strongly Nonlinear Concept for Geophysical Turbulence

   https://doi.org/doi:10.3390/fluids2030036  

   Polzin, Kurt L.; Lvov, Yuri (June 2017, Fluids)

   Abstract: There is no theoretical underpinning that successfully explains how turbulent mixing is fed by wave breaking associated with nonlinear wave-wave interactions in the background oceanic internal wavefield. We address this conundrum using one-dimensional ray tracing simulations to investigate interactions between high frequency internal waves and inertial oscillations in the extreme scale separated limit known as “Induced Diffusion”. Here, estimates of phase locking are used to define a resonant process (a resonant well) and a non-resonant process that results in stochastic jumps. The small amplitude limit consists of jumps that are small compared to the scale of the resonant well. The ray tracing simulations are used to estimate the first and second moments of a wave packet’s vertical wavenumber as it evolves from an initial condition. These moments are compared with predictions obtained from the diffusive approximation to a self-consistent kinetic equation derived in the ‘Direct Interaction Approximation’. Results indicate that the first and second moments of the two systems evolve in a nearly identical manner when the inertial field has amplitudes an order of magnitude smaller than oceanic values. At realistic (oceanic) amplitudes, though, the second moment estimated from the ray tracing simulations is inhibited. The transition is explained by the stochastic jumps obtaining the characteristic size of the resonant well. We interpret this transition as an adiabatic ‘saturation’ process which changes the nominal background wavefield from supporting no mixing to the point where that background wavefield defines the normalization for oceanic mixing models. 

   more » « less
4. Fault Detection Effectiveness of Metamorphic Relations Developed for Testing Supervised Classifiers

   https://doi.org/10.1109/AITest.2019.00019  

   Saha, Prashanta; Kanewala, Upulee (April 2019, 2019 IEEE International Conference On Artificial Intelligence Testing (AITest))

   In machine learning, supervised classifiers are used to obtain predictions for unlabeled data by inferring prediction functions using labeled data. Supervised classifiers are widely applied in domains such as computational biology, computational physics and healthcare to make critical decisions. However, it is often hard to test supervised classifiers since the expected answers are unknown. This is commonly known as the oracle problem and metamorphic testing (MT) has been used to test such programs. In MT, metamorphic relations (MRs) are developed from intrinsic characteristics of the software under test (SUT). These MRs are used to generate test data and to verify the correctness of the test results without the presence of a test oracle. Effectiveness of MT heavily depends on the MRs used for testing. In this paper we have conducted an extensive empirical study to evaluate the fault detection effectiveness of MRs that have been used in multiple previous studies to test supervised classifiers. Our study uses a total of 709 reachable mutants generated by multiple mutation engines and uses data sets with varying characteristics to test the SUT. Our results reveal that only 14.8% of these mutants are detected using the MRs and that the fault detection effectiveness of these MRs do not scale with the increased number of mutants when compared to what was reported in previous studies. 

   more » « less
5. Keeping up with the genomes: efficient learning of our increasing knowledge of the tree of life

   https://doi.org/10.1186/s12859-020-03744-7  

   Zhao, Zhengqiao; Cristian, Alexandru; Rosen, Gail (December 2020, BMC Bioinformatics)

   null (Ed.)

   Abstract Background It is a computational challenge for current metagenomic classifiers to keep up with the pace of training data generated from genome sequencing projects, such as the exponentially-growing NCBI RefSeq bacterial genome database. When new reference sequences are added to training data, statically trained classifiers must be rerun on all data, resulting in a highly inefficient process. The rich literature of “incremental learning” addresses the need to update an existing classifier to accommodate new data without sacrificing much accuracy compared to retraining the classifier with all data. Results We demonstrate how classification improves over time by incrementally training a classifier on progressive RefSeq snapshots and testing it on: (a) all known current genomes (as a ground truth set) and (b) a real experimental metagenomic gut sample. We demonstrate that as a classifier model’s knowledge of genomes grows, classification accuracy increases. The proof-of-concept naïve Bayes implementation, when updated yearly, now runs in 1/4 t h of the non-incremental time with no accuracy loss. Conclusions It is evident that classification improves by having the most current knowledge at its disposal. Therefore, it is of utmost importance to make classifiers computationally tractable to keep up with the data deluge. The incremental learning classifier can be efficiently updated without the cost of reprocessing nor the access to the existing database and therefore save storage as well as computation resources. 

   more » « less