Saavan Patel, Philip Canoza, Adhiraj Datar, Steven Lu, Chirag Garg, Sayeef Salahuddin
New computing paradigms are required to solve the most challenging computational problems where no exact polynomial time solution exists.Probabilistic Ising Accelerators has gained promise on these problems with the ability to model complex probability distributions and find ground states of intractable problems. In this context, we have demonstrated the Parallel Asynchronous Stochastic Sampler (PASS), the first fully on-chip integrated, asynchronous, probabilistic accelerator that takes advantage of the intrinsic fine-grained parallelism of the Ising Model and built in state of the art 14nm CMOS FinFET technology. We have demonstrated broad applicability of this accelerator on problems ranging from Combinatorial Optimization, Neural Simulation, to Machine Learning along with up to $23,000$x energy to solution improvement compared to CPUs on probabilistic problems.
要解决没有精确多项式时间解决方案的最具挑战性的计算问题，就必须采用新的计算范式。概率伊辛加速器能够模拟复杂的概率分布并找到棘手问题的基态，因此在这些问题上大有可为。在此背景下，我们展示了并行异步随机取样器 (PASS)，这是首个完全集成在芯片上的异步概率加速器，它利用了伊辛模型内在的细粒度并行性，并采用最先进的 14nm CMOS FinFET 技术。我们已经证明了该加速器在组合优化、神经仿真和机器学习等问题上的广泛适用性，与 CPU 相比，在概率问题上的能效提高了 23,000 美元。
{"title":"PASS: An Asynchronous Probabilistic Processor for Next Generation Intelligence","authors":"Saavan Patel, Philip Canoza, Adhiraj Datar, Steven Lu, Chirag Garg, Sayeef Salahuddin","doi":"arxiv-2409.10325","DOIUrl":"https://doi.org/arxiv-2409.10325","url":null,"abstract":"New computing paradigms are required to solve the most challenging\u0000computational problems where no exact polynomial time solution\u0000exists.Probabilistic Ising Accelerators has gained promise on these problems\u0000with the ability to model complex probability distributions and find ground\u0000states of intractable problems. In this context, we have demonstrated the\u0000Parallel Asynchronous Stochastic Sampler (PASS), the first fully on-chip\u0000integrated, asynchronous, probabilistic accelerator that takes advantage of the\u0000intrinsic fine-grained parallelism of the Ising Model and built in state of the\u0000art 14nm CMOS FinFET technology. We have demonstrated broad applicability of\u0000this accelerator on problems ranging from Combinatorial Optimization, Neural\u0000Simulation, to Machine Learning along with up to $23,000$x energy to solution\u0000improvement compared to CPUs on probabilistic problems.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"54 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142253351","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
With nearly two billion stars observed and their corresponding astrometric parameters evaluated in the recent Gaia mission, the number of astrometric binary candidates have risen significantly. Due to the surplus of astrometric data, the current computational methods employed to inspect these astrometric binary candidates are both computationally expensive and cannot be executed in a reasonable time frame. In light of this, a machine learning (ML) technique to automatically classify whether a set of stars belong to an astrometric binary pair via an artificial neural network (ANN) is proposed. Using data from Gaia DR3, the ANN was trained and tested on 1.5 million highly probable true and visual binaries, considering the proper motions, parallaxes, and angular and physical separations as features. The ANN achieves high classification scores, with an accuracy of 99.3%, a precision rate of 0.988, a recall rate of 0.991, and an AUC of 0.999, indicating that the utilized ML technique is a highly effective method for classifying astrometric binaries. Thus, the proposed ANN is a promising alternative to the existing methods for the classification of astrometric binaries.
{"title":"Astrometric Binary Classification Via Artificial Neural Networks","authors":"Joe Smith","doi":"arxiv-2409.09563","DOIUrl":"https://doi.org/arxiv-2409.09563","url":null,"abstract":"With nearly two billion stars observed and their corresponding astrometric\u0000parameters evaluated in the recent Gaia mission, the number of astrometric\u0000binary candidates have risen significantly. Due to the surplus of astrometric\u0000data, the current computational methods employed to inspect these astrometric\u0000binary candidates are both computationally expensive and cannot be executed in\u0000a reasonable time frame. In light of this, a machine learning (ML) technique to\u0000automatically classify whether a set of stars belong to an astrometric binary\u0000pair via an artificial neural network (ANN) is proposed. Using data from Gaia\u0000DR3, the ANN was trained and tested on 1.5 million highly probable true and\u0000visual binaries, considering the proper motions, parallaxes, and angular and\u0000physical separations as features. The ANN achieves high classification scores,\u0000with an accuracy of 99.3%, a precision rate of 0.988, a recall rate of 0.991,\u0000and an AUC of 0.999, indicating that the utilized ML technique is a highly\u0000effective method for classifying astrometric binaries. Thus, the proposed ANN\u0000is a promising alternative to the existing methods for the classification of\u0000astrometric binaries.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"66 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142253353","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
XENON Collaboration, E. Aprile, J. Aalbers, K. Abe, S. Ahmed Maouloud, L. Althueser, B. Andrieu, E. Angelino, J. R. Angevaare, D. Antón Martin, F. Arneodo, L. Baudis, M. Bazyk, L. Bellagamba, R. Biondi, A. Bismark, K. Boese, A. Brown, G. Bruno, R. Budnik, J. M. R. Cardoso, A. P. Cimental Chávez, A. P. Colijn, J. Conrad, J. J. Cuenca-García, V. D'Andrea, L. C. Daniel Garcia, M. P. Decowski, A. Deisting, C. Di Donato, P. Di Gangi, S. Diglio, K. Eitel, A. Elykov, A. D. Ferella, C. Ferrari, H. Fischer, T. Flehmke, M. Flierman, W. Fulgione, C. Fuselli, P. Gaemers, R. Gaior, M. Galloway, F. Gao, S. Ghosh, R. Giacomobono, R. Glade-Beucke, L. Grandi, J. Grigat, H. Guan, M. Guida, P. Gyoergy, R. Hammann, A. Higuera, C. Hils, L. Hoetzsch, N. F. Hood, M. Iacovacci, Y. Itow, J. Jakob, F. Joerg, Y. Kaminaga, M. Kara, P. Kavrigin, S. Kazama, M. Kobayashi, D. Koke, A. Kopec, F. Kuger, H. Landsman, R. F. Lang, L. Levinson, I. Li, S. Li, S. Liang, Y. -T. Lin, S. Lindemann, M. Lindner, K. Liu, J. Loizeau, F. Lombardi, J. Long, J. A. M. Lopes, T. Luce, Y. Ma, C. Macolino, J. Mahlstedt, A. Mancuso, L. Manenti, F. Marignetti, T. Marrodán Undagoitia, K. Martens, J. Masbou, E. Masson, S. Mastroianni, A. Melchiorre, J. Merz, M. Messina, A. Michael, K. Miuchi, A. Molinario, S. Moriyama, K. Morå, Y. Mosbacher, M. Murra, J. Müller, K. Ni, U. Oberlack, B. Paetsch, Y. Pan, Q. Pellegrini, R. Peres, C. Peters, J. Pienaar, M. Pierre, G. Plante, T. R. Pollmann, L. Principe, J. Qi, J. Qin, D. Ramírez García, M. Rajado, R. Singh, L. Sanchez, J. M. F. dos Santos, I. Sarnoff, G. Sartorelli, J. Schreiner, D. Schulte, P. Schulte, H. Schulze Eißing, M. Schumann, L. Scotto Lavina, M. Selvi, F. Semeria, P. Shagin, S. Shi, J. Shi, M. Silva, H. Simgen, A. Takeda, P. -L. Tan, A. Terliuk, D. Thers, F. Toschi, G. Trinchero, C. D. Tunnell, F. Tönnies, K. Valerius, S. Vecchi, S. Vetter, F. I. Villazon Solar, G. Volta, C. Weinheimer, M. Weiss, D. Wenz, C. Wittweg, V. H. S. Wu, Y. Xing, D. Xu, Z. Xu, M. Yamashita, L. Yang, J. Ye, L. Yuan, G. Zavattini, M. Zhong
The XENONnT experiment, located at the INFN Laboratori Nazionali del Gran Sasso, Italy, features a 5.9 tonne liquid xenon time projection chamber surrounded by an instrumented neutron veto, all of which is housed within a muon veto water tank. Due to extensive shielding and advanced purification to mitigate natural radioactivity, an exceptionally low background level of (15.8 $pm$ 1.3) events/(tonne$cdot$year$cdot$keV) in the (1, 30) keV region is reached in the inner part of the TPC. XENONnT is thus sensitive to a wide range of rare phenomena related to Dark Matter and Neutrino interactions, both within and beyond the Standard Model of particle physics, with a focus on the direct detection of Dark Matter in the form of weakly interacting massive particles (WIMPs). From May 2021 to December 2021, XENONnT accumulated data in rare-event search mode with a total exposure of one tonne $cdot$ year. This paper provides a detailed description of the signal reconstruction methods, event selection procedure, and detector response calibration, as well as an overview of the detector performance in this time frame. This work establishes the foundational framework for the `blind analysis' methodology we are using when reporting XENONnT physics results.
{"title":"XENONnT Analysis: Signal Reconstruction, Calibration and Event Selection","authors":"XENON Collaboration, E. Aprile, J. Aalbers, K. Abe, S. Ahmed Maouloud, L. Althueser, B. Andrieu, E. Angelino, J. R. Angevaare, D. Antón Martin, F. Arneodo, L. Baudis, M. Bazyk, L. Bellagamba, R. Biondi, A. Bismark, K. Boese, A. Brown, G. Bruno, R. Budnik, J. M. R. Cardoso, A. P. Cimental Chávez, A. P. Colijn, J. Conrad, J. J. Cuenca-García, V. D'Andrea, L. C. Daniel Garcia, M. P. Decowski, A. Deisting, C. Di Donato, P. Di Gangi, S. Diglio, K. Eitel, A. Elykov, A. D. Ferella, C. Ferrari, H. Fischer, T. Flehmke, M. Flierman, W. Fulgione, C. Fuselli, P. Gaemers, R. Gaior, M. Galloway, F. Gao, S. Ghosh, R. Giacomobono, R. Glade-Beucke, L. Grandi, J. Grigat, H. Guan, M. Guida, P. Gyoergy, R. Hammann, A. Higuera, C. Hils, L. Hoetzsch, N. F. Hood, M. Iacovacci, Y. Itow, J. Jakob, F. Joerg, Y. Kaminaga, M. Kara, P. Kavrigin, S. Kazama, M. Kobayashi, D. Koke, A. Kopec, F. Kuger, H. Landsman, R. F. Lang, L. Levinson, I. Li, S. Li, S. Liang, Y. -T. Lin, S. Lindemann, M. Lindner, K. Liu, J. Loizeau, F. Lombardi, J. Long, J. A. M. Lopes, T. Luce, Y. Ma, C. Macolino, J. Mahlstedt, A. Mancuso, L. Manenti, F. Marignetti, T. Marrodán Undagoitia, K. Martens, J. Masbou, E. Masson, S. Mastroianni, A. Melchiorre, J. Merz, M. Messina, A. Michael, K. Miuchi, A. Molinario, S. Moriyama, K. Morå, Y. Mosbacher, M. Murra, J. Müller, K. Ni, U. Oberlack, B. Paetsch, Y. Pan, Q. Pellegrini, R. Peres, C. Peters, J. Pienaar, M. Pierre, G. Plante, T. R. Pollmann, L. Principe, J. Qi, J. Qin, D. Ramírez García, M. Rajado, R. Singh, L. Sanchez, J. M. F. dos Santos, I. Sarnoff, G. Sartorelli, J. Schreiner, D. Schulte, P. Schulte, H. Schulze Eißing, M. Schumann, L. Scotto Lavina, M. Selvi, F. Semeria, P. Shagin, S. Shi, J. Shi, M. Silva, H. Simgen, A. Takeda, P. -L. Tan, A. Terliuk, D. Thers, F. Toschi, G. Trinchero, C. D. Tunnell, F. Tönnies, K. Valerius, S. Vecchi, S. Vetter, F. I. Villazon Solar, G. Volta, C. Weinheimer, M. Weiss, D. Wenz, C. Wittweg, V. H. S. Wu, Y. Xing, D. Xu, Z. Xu, M. Yamashita, L. Yang, J. Ye, L. Yuan, G. Zavattini, M. Zhong","doi":"arxiv-2409.08778","DOIUrl":"https://doi.org/arxiv-2409.08778","url":null,"abstract":"The XENONnT experiment, located at the INFN Laboratori Nazionali del Gran\u0000Sasso, Italy, features a 5.9 tonne liquid xenon time projection chamber\u0000surrounded by an instrumented neutron veto, all of which is housed within a\u0000muon veto water tank. Due to extensive shielding and advanced purification to\u0000mitigate natural radioactivity, an exceptionally low background level of (15.8\u0000$pm$ 1.3) events/(tonne$cdot$year$cdot$keV) in the (1, 30) keV region is\u0000reached in the inner part of the TPC. XENONnT is thus sensitive to a wide range\u0000of rare phenomena related to Dark Matter and Neutrino interactions, both within\u0000and beyond the Standard Model of particle physics, with a focus on the direct\u0000detection of Dark Matter in the form of weakly interacting massive particles\u0000(WIMPs). From May 2021 to December 2021, XENONnT accumulated data in rare-event\u0000search mode with a total exposure of one tonne $cdot$ year. This paper\u0000provides a detailed description of the signal reconstruction methods, event\u0000selection procedure, and detector response calibration, as well as an overview\u0000of the detector performance in this time frame. This work establishes the\u0000foundational framework for the `blind analysis' methodology we are using when\u0000reporting XENONnT physics results.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142253354","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The use of machine learning approaches continues to have many benefits in experimental nuclear and particle physics. One common issue is generating training data which is sufficiently realistic to give reliable results. Here we advocate using real experimental data as the source of training data and demonstrate how one might subtract background contributions through the use of probabilistic weights which can be readily applied to training data. The sPlot formalism is a common tool used to isolate distributions from different sources. However, negative sWeights produced by the sPlot technique can lead to issues in training and poor predictive power. This article demonstrates how density ratio estimation can be applied to convert sWeights to event probabilities, which we call drWeights. The drWeights can then be applied to produce the distributions of interest and are consistent with direct use of the sWeights. This article will also show how decision trees are particular well suited to converting sWeights, with the benefit of fast prediction rates and adaptability to aspects of the experimental data such as data sample size and proportions of different event sources. We also show that a double density ratio approach where the initial drWeights are reweighted by an additional classifier gives substantially better results.
{"title":"Converting sWeights to Probabilities with Density Ratios","authors":"D. I. Glazier, R. Tyson","doi":"arxiv-2409.08183","DOIUrl":"https://doi.org/arxiv-2409.08183","url":null,"abstract":"The use of machine learning approaches continues to have many benefits in\u0000experimental nuclear and particle physics. One common issue is generating\u0000training data which is sufficiently realistic to give reliable results. Here we\u0000advocate using real experimental data as the source of training data and\u0000demonstrate how one might subtract background contributions through the use of\u0000probabilistic weights which can be readily applied to training data. The sPlot\u0000formalism is a common tool used to isolate distributions from different\u0000sources. However, negative sWeights produced by the sPlot technique can lead to\u0000issues in training and poor predictive power. This article demonstrates how\u0000density ratio estimation can be applied to convert sWeights to event\u0000probabilities, which we call drWeights. The drWeights can then be applied to\u0000produce the distributions of interest and are consistent with direct use of the\u0000sWeights. This article will also show how decision trees are particular well\u0000suited to converting sWeights, with the benefit of fast prediction rates and\u0000adaptability to aspects of the experimental data such as data sample size and\u0000proportions of different event sources. We also show that a double density\u0000ratio approach where the initial drWeights are reweighted by an additional\u0000classifier gives substantially better results.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"14 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142179395","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The analysis of event time series is in general challenging. Most time series analysis tools are limited for the analysis of this kind of data. Recurrence analysis, a powerful concept from nonlinear time series analysis, provides several opportunities to work with event data and even for the most challenging task of comparing event time series with continuous time series. Here, the basic concept is introduced, the challenges are discussed, and the future perspectives are summarised.
{"title":"Challenges and perspectives in recurrence analyses of event time series","authors":"Norbert Marwan","doi":"arxiv-2409.08398","DOIUrl":"https://doi.org/arxiv-2409.08398","url":null,"abstract":"The analysis of event time series is in general challenging. Most time series\u0000analysis tools are limited for the analysis of this kind of data. Recurrence\u0000analysis, a powerful concept from nonlinear time series analysis, provides\u0000several opportunities to work with event data and even for the most challenging\u0000task of comparing event time series with continuous time series. Here, the\u0000basic concept is introduced, the challenges are discussed, and the future\u0000perspectives are summarised.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"6 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142253355","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Michael Parzer, Alexander Riss, Fabian Garmroudi, Johannes de Boor, Takao Mori, Ernst Bauer
Linking the fundamental physics of band structure and scattering theory with macroscopic features such as measurable bulk thermoelectric transport properties is indispensable to a thorough understanding of transport phenomena and ensures more targeted and efficient experimental research. Here, we introduce SeeBand, a highly efficient and interactive fitting tool based on Boltzmann transport theory. A fully integrated user interface and visualization tool enable real-time comparison and connection between the electronic band structure (EBS) and microscopic transport properties. It allows simultaneous analysis of data for the Seebeck coefficient $S$, resistivity $rho$ and Hall coefficient $R_text{H}$ to identify suitable EBS models and extract the underlying microscopic material parameters and additional information from the model. Crucially, the EBS can be obtained by directly fitting the temperature-dependent properties of a single sample, which goes beyond previous approaches that look into doping dependencies. Finally, the combination of neural-network-assisted initial guesses and an efficient subsequent fitting routine allows for a rapid processing of big datasets, facilitating high-throughput analyses to identify underlying, yet undiscovered dependencies, thereby guiding material design.
{"title":"SeeBand: A highly efficient, interactive tool for analyzing electronic transport data","authors":"Michael Parzer, Alexander Riss, Fabian Garmroudi, Johannes de Boor, Takao Mori, Ernst Bauer","doi":"arxiv-2409.06261","DOIUrl":"https://doi.org/arxiv-2409.06261","url":null,"abstract":"Linking the fundamental physics of band structure and scattering theory with\u0000macroscopic features such as measurable bulk thermoelectric transport\u0000properties is indispensable to a thorough understanding of transport phenomena\u0000and ensures more targeted and efficient experimental research. Here, we\u0000introduce SeeBand, a highly efficient and interactive fitting tool based on\u0000Boltzmann transport theory. A fully integrated user interface and visualization\u0000tool enable real-time comparison and connection between the electronic band\u0000structure (EBS) and microscopic transport properties. It allows simultaneous\u0000analysis of data for the Seebeck coefficient $S$, resistivity $rho$ and Hall\u0000coefficient $R_text{H}$ to identify suitable EBS models and extract the\u0000underlying microscopic material parameters and additional information from the\u0000model. Crucially, the EBS can be obtained by directly fitting the\u0000temperature-dependent properties of a single sample, which goes beyond previous\u0000approaches that look into doping dependencies. Finally, the combination of\u0000neural-network-assisted initial guesses and an efficient subsequent fitting\u0000routine allows for a rapid processing of big datasets, facilitating\u0000high-throughput analyses to identify underlying, yet undiscovered dependencies,\u0000thereby guiding material design.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"15 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142179396","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The shortcomings of the Standard Model (SM) motivate its extension to accommodate new expected phenomena, such as dark matter and neutrino masses. However, such extensions are generally more complex due to the presence of a larger number of free parameters as well as additional phenomenology. Understanding how current theoretical and experimental constraints, individually and collectively, affect the parameter spaces of new models is of utmost importance in achieving testable predictions and targeted model-building that aims to solve certain issues. We present a comprehensive approach of using Deep Learning (DL) for the multi-label classification (MLC) of theoretical and experimental limits on the two-Higgs doublet model augmented by a real singlet (N2HDM), as a representative case. This approach can be generalized to any extension beyond the SM.
标准模型（SM）的缺陷促使其扩展以适应新的预期现象，如暗物质和中微子质量。然而，由于存在更多的自由参数以及额外的现象学，这种扩展通常更为复杂。了解当前的理论和实验约束如何单独或集体地影响新模型的参数空间，对于实现可检验的预测和旨在解决某些问题的有针对性的模型构建至关重要。我们提出了一种综合方法，即使用深度学习（Deep Learning，DL）对以实单子增强的双希格斯双子模型（N2HDM）为代表的理论和实验限制进行多标签分类（MLC）。这种方法可以推广到 SM 以外的任何扩展。
{"title":"Multi-label Classification of Parameter Constraints in BSM Extensions using Deep Learning","authors":"Maien Binjonaid","doi":"arxiv-2409.05453","DOIUrl":"https://doi.org/arxiv-2409.05453","url":null,"abstract":"The shortcomings of the Standard Model (SM) motivate its extension to\u0000accommodate new expected phenomena, such as dark matter and neutrino masses.\u0000However, such extensions are generally more complex due to the presence of a\u0000larger number of free parameters as well as additional phenomenology.\u0000Understanding how current theoretical and experimental constraints,\u0000individually and collectively, affect the parameter spaces of new models is of\u0000utmost importance in achieving testable predictions and targeted model-building\u0000that aims to solve certain issues. We present a comprehensive approach of using\u0000Deep Learning (DL) for the multi-label classification (MLC) of theoretical and\u0000experimental limits on the two-Higgs doublet model augmented by a real singlet\u0000(N2HDM), as a representative case. This approach can be generalized to any\u0000extension beyond the SM.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"32 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142179266","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The performance of Quantum Autoencoders (QAEs) in anomaly detection tasks is critically dependent on the choice of data embedding and ansatz design. This study explores the effects of three data embedding techniques, data re-uploading, parallel embedding, and alternate embedding, on the representability and effectiveness of QAEs in detecting anomalies. Our findings reveal that even with relatively simple variational circuits, enhanced data embedding strategies can substantially improve anomaly detection accuracy and the representability of underlying data across different datasets. Starting with toy examples featuring low-dimensional data, we visually demonstrate the effect of different embedding techniques on the representability of the model. We then extend our analysis to complex, higher-dimensional datasets, highlighting the significant impact of embedding methods on QAE performance.
{"title":"The role of data embedding in quantum autoencoders for improved anomaly detection","authors":"Jack Y. Araz, Michael Spannowsky","doi":"arxiv-2409.04519","DOIUrl":"https://doi.org/arxiv-2409.04519","url":null,"abstract":"The performance of Quantum Autoencoders (QAEs) in anomaly detection tasks is\u0000critically dependent on the choice of data embedding and ansatz design. This\u0000study explores the effects of three data embedding techniques, data\u0000re-uploading, parallel embedding, and alternate embedding, on the\u0000representability and effectiveness of QAEs in detecting anomalies. Our findings\u0000reveal that even with relatively simple variational circuits, enhanced data\u0000embedding strategies can substantially improve anomaly detection accuracy and\u0000the representability of underlying data across different datasets. Starting\u0000with toy examples featuring low-dimensional data, we visually demonstrate the\u0000effect of different embedding techniques on the representability of the model.\u0000We then extend our analysis to complex, higher-dimensional datasets,\u0000highlighting the significant impact of embedding methods on QAE performance.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"17 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142179397","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The last decade has witnessed a number of important and exciting developments that had been achieved for improving recurrence plot based data analysis and to widen its application potential. We will give a brief overview about important and innovative developments, such as computational improvements, alternative recurrence definitions (event-like, multiscale, heterogeneous, and spatio-temporal recurrences) and ideas for parameter selection, theoretical considerations of recurrence quantification measures, new recurrence quantifiers (e.g., for transition detection and causality detection), and correction schemes. New perspectives have recently been opened by combining recurrence plots with machine learning. We finally show open questions and perspectives for futures directions of methodical research.
{"title":"Trends in recurrence analysis of dynamical systems","authors":"Norbert Marwan, K. Hauke Kraemer","doi":"arxiv-2409.04110","DOIUrl":"https://doi.org/arxiv-2409.04110","url":null,"abstract":"The last decade has witnessed a number of important and exciting developments\u0000that had been achieved for improving recurrence plot based data analysis and to\u0000widen its application potential. We will give a brief overview about important\u0000and innovative developments, such as computational improvements, alternative\u0000recurrence definitions (event-like, multiscale, heterogeneous, and\u0000spatio-temporal recurrences) and ideas for parameter selection, theoretical\u0000considerations of recurrence quantification measures, new recurrence\u0000quantifiers (e.g., for transition detection and causality detection), and\u0000correction schemes. New perspectives have recently been opened by combining\u0000recurrence plots with machine learning. We finally show open questions and\u0000perspectives for futures directions of methodical research.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"14 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142179398","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Taichi YamamotoThe University of Tokyo, Hiroya NakaoTokyo Institute of Technology, Ryota KobayashiThe University of Tokyo
Rhythmic activity commonly observed in biological systems, occurring from the cellular level to the organismic level, is typically modeled as limit cycle oscillators. The phase reduction theory serves as a useful analytical framework for elucidating the synchronization mechanism of these oscillators. Essentially, this theory describes the dynamics of a multi-dimensional nonlinear oscillator using a single variable phase model. In order to understand and control the rhythmic phenomena in the real world, it is crucial to estimate the asymptotic phase from the observed data. In this study, we propose a new method, Gaussian Process Phase Interpolation (GPPI), for estimating the asymptotic phase from time series data. The GPPI method first evaluates the asymptotic phase on the limit cycle and subsequently estimates the asymptotic phase outside the limit cycle employing Gaussian process regression. Thanks to the high expressive power of Gaussian processes, the GPPI is capable of capturing a variety of functions. Notably, the GPPI is easily applicable even when the dimension of the system increases. The performance of the GPPI is tested by using simulation data from the Stuart-Landau oscillator and the Hodgkin-Huxley oscillator. The results demonstrate that the GPPI can accurately estimate the asymptotic phase even in the presence of high observation noise and strong nonlinearity. Additionally, the GPPI is demonstrated as an effective tool for data-driven phase control of a Hodgkin-Huxley oscillator. Thus, the proposed GPPI will facilitate the data-driven modeling of the limit cycle oscillators.
{"title":"Gaussian Process Phase Interpolation for estimating the asymptotic phase of a limit cycle oscillator from time series data","authors":"Taichi YamamotoThe University of Tokyo, Hiroya NakaoTokyo Institute of Technology, Ryota KobayashiThe University of Tokyo","doi":"arxiv-2409.03290","DOIUrl":"https://doi.org/arxiv-2409.03290","url":null,"abstract":"Rhythmic activity commonly observed in biological systems, occurring from the\u0000cellular level to the organismic level, is typically modeled as limit cycle\u0000oscillators. The phase reduction theory serves as a useful analytical framework\u0000for elucidating the synchronization mechanism of these oscillators.\u0000Essentially, this theory describes the dynamics of a multi-dimensional\u0000nonlinear oscillator using a single variable phase model. In order to\u0000understand and control the rhythmic phenomena in the real world, it is crucial\u0000to estimate the asymptotic phase from the observed data. In this study, we\u0000propose a new method, Gaussian Process Phase Interpolation (GPPI), for\u0000estimating the asymptotic phase from time series data. The GPPI method first\u0000evaluates the asymptotic phase on the limit cycle and subsequently estimates\u0000the asymptotic phase outside the limit cycle employing Gaussian process\u0000regression. Thanks to the high expressive power of Gaussian processes, the GPPI\u0000is capable of capturing a variety of functions. Notably, the GPPI is easily\u0000applicable even when the dimension of the system increases. The performance of\u0000the GPPI is tested by using simulation data from the Stuart-Landau oscillator\u0000and the Hodgkin-Huxley oscillator. The results demonstrate that the GPPI can\u0000accurately estimate the asymptotic phase even in the presence of high\u0000observation noise and strong nonlinearity. Additionally, the GPPI is\u0000demonstrated as an effective tool for data-driven phase control of a\u0000Hodgkin-Huxley oscillator. Thus, the proposed GPPI will facilitate the\u0000data-driven modeling of the limit cycle oscillators.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"18 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142179399","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：