-
Heavy neutral leptons and top quarks in effective field theory
Authors:
Rebeca Beltrán,
Giovanna Cottin,
Julian Günther,
Martin Hirsch,
Arsenii Titov,
Zeren Simon Wang
Abstract:
We study the phenomenology of heavy neutral leptons (HNLs) at the LHC in effective field theory, concentrating on $d=6$ operators with top quarks. Depending on the operator choice and HNL mass, the HNLs will be produced either from proton-proton collisions in association with a single top, or via non-standard decays of top quarks. For long-lived HNLs we estimate the sensitivity reach of different…
▽ More
We study the phenomenology of heavy neutral leptons (HNLs) at the LHC in effective field theory, concentrating on $d=6$ operators with top quarks. Depending on the operator choice and HNL mass, the HNLs will be produced either from proton-proton collisions in association with a single top, or via non-standard decays of top quarks. For long-lived HNLs we estimate the sensitivity reach of different detectors to various operators with top quarks and the HNLs for the high-luminosity phase of the LHC. For certain operators, ATLAS and some far detectors (MATHUSLA and ANUBIS) will be able to probe the associated new-physics scale as large as 12 TeV and 4.5 TeV, respectively, covering complementary HNL-mass ranges.
△ Less
Submitted 29 April, 2025; v1 submitted 15 January, 2025;
originally announced January 2025.
-
Towards Efficient and Scalable Training of Differentially Private Deep Learning
Authors:
Sebastian Rodriguez Beltran,
Marlon Tobaben,
Joonas Jälkö,
Niki Loppi,
Antti Honkela
Abstract:
Differentially private stochastic gradient descent (DP-SGD) is the standard algorithm for training machine learning models under differential privacy (DP). The most common DP-SGD privacy accountants rely on Poisson subsampling for ensuring the theoretical DP guarantees. Implementing computationally efficient DP-SGD with Poisson subsampling is not trivial, which leads to many implementations ignori…
▽ More
Differentially private stochastic gradient descent (DP-SGD) is the standard algorithm for training machine learning models under differential privacy (DP). The most common DP-SGD privacy accountants rely on Poisson subsampling for ensuring the theoretical DP guarantees. Implementing computationally efficient DP-SGD with Poisson subsampling is not trivial, which leads to many implementations ignoring this requirement. We conduct a comprehensive empirical study to quantify the computational cost of training deep learning models under DP given the requirement of Poisson subsampling, by re-implementing efficient methods using Poisson subsampling and benchmarking them. We find that using the naive implementation DP-SGD with Opacus in PyTorch has between 2.6 and 8 times lower throughput of processed training examples per second than SGD. However, efficient gradient clipping implementations with e.g. Ghost Clipping can roughly halve this cost. We propose alternative computationally efficient ways of implementing DP-SGD with JAX that are using Poisson subsampling and achieve only around 1.2 times lower throughput than SGD based on PyTorch. We highlight important implementation considerations with JAX. Finally, we study the scaling behaviour using up to 80 GPUs and find that DP-SGD scales better than SGD. We share our re-implementations using Poisson subsampling at https://github.com/DPBayes/Towards-Efficient-Scalable-Training-DP-DL.
△ Less
Submitted 27 October, 2024; v1 submitted 25 June, 2024;
originally announced June 2024.
-
Probing Heavy Neutrino Magnetic Moments at the LHC using Long-Lived Particle Searches
Authors:
Rebeca Beltrán,
Patrick D. Bolton,
Frank F. Deppisch,
Chandan Hati,
Martin Hirsch
Abstract:
We explore long-lived particle (LLP) searches using non-pointing photons at the LHC as a probe for sterile-to-sterile and active-to-sterile transition magnetic dipole moments of sterile neutrinos. We consider heavy sterile neutrinos with masses ranging from a few~GeV to several hundreds of GeV. We discuss transition magnetic dipole moments using the Standard Model effective field theory and low-en…
▽ More
We explore long-lived particle (LLP) searches using non-pointing photons at the LHC as a probe for sterile-to-sterile and active-to-sterile transition magnetic dipole moments of sterile neutrinos. We consider heavy sterile neutrinos with masses ranging from a few~GeV to several hundreds of GeV. We discuss transition magnetic dipole moments using the Standard Model effective field theory and low-energy effective field theory extended by sterile neutrinos ($N_R$SMEFT and $N_R$LEFT) and also provide a simplified UV-complete model example. LLP searches at the LHC using non-pointing photons will probe sterile-to-sterile dipole moments two orders of magnitude below the current best constraints from LEP, while an unprecedented sensitivity to sterile neutrino mass of about 700 GeV is expected for active-to-sterile dipole moments. For the UV model example with one-loop transition magnetic moments, the searches for charged lepton flavour violating processes in synergy with LLP searches at the LHC can probe new physics at several TeV mass scales and provide valuable insights into the lepton flavour structure of new physics couplings.
△ Less
Submitted 28 August, 2024; v1 submitted 14 May, 2024;
originally announced May 2024.
-
Heavy neutral leptons from kaons in effective field theory
Authors:
Rebeca Beltrán,
Julian Günther,
Martin Hirsch,
Arsenii Titov,
Zeren Simon Wang
Abstract:
In the framework of the low-energy effective theory containing in addition to the Standard Model fields heavy neutral leptons (HNLs), we compute the decay rates of neutral and charged kaons into HNLs. We consider both lepton-number-conserving and lepton-number-violating four-fermion operators, taking into account also the contribution of active-heavy neutrino mixing. Assuming that the produced HNL…
▽ More
In the framework of the low-energy effective theory containing in addition to the Standard Model fields heavy neutral leptons (HNLs), we compute the decay rates of neutral and charged kaons into HNLs. We consider both lepton-number-conserving and lepton-number-violating four-fermion operators, taking into account also the contribution of active-heavy neutrino mixing. Assuming that the produced HNLs are long-lived, we perform simulations and calculate the sensitivities of future long-lived-particle (LLP) detectors at the high-luminosity LHC as well as the near detector of the Deep Underground Neutrino Experiment (DUNE-ND) to the considered scenario. When applicable, we also recast the existing bounds on the minimal mixing case obtained by NA62, T2K, and PS191. Our findings show that while the future LHC LLP detectors can probe currently allowed parameter space only in certain benchmark scenarios, DUNE-ND should be sensitive to parameter space beyond the current bounds in almost all the benchmark scenarios and for some of the effective operators considered it can even probe new-physics scales in excess of 3000 TeV.
△ Less
Submitted 22 May, 2024; v1 submitted 20 September, 2023;
originally announced September 2023.
-
Tree-level UV completions for $N_R$SMEFT $d=6$ and $d=7$ operators
Authors:
Rebeca Beltrán,
Ricardo Cepedello,
Martin Hirsch
Abstract:
We study ultra-violet completions for operators in standard model effective field theory extended with right-handed neutrinos ($N_R$SMEFT). Using a diagrammatic method, we generate systematically lists of possible tree-level completions involving scalars, fermions or vectors for all operators at $d=6$ and $d=7$, which contain at least one right-handed neutrino. We compare our lists of possible UV…
▽ More
We study ultra-violet completions for operators in standard model effective field theory extended with right-handed neutrinos ($N_R$SMEFT). Using a diagrammatic method, we generate systematically lists of possible tree-level completions involving scalars, fermions or vectors for all operators at $d=6$ and $d=7$, which contain at least one right-handed neutrino. We compare our lists of possible UV models to the ones found for pure SMEFT. We also discuss how the observation of LNV processes via $N_R$SMEFT operators at the LHC can be related to Majorana neutrino masses of the standard model neutrinos.
△ Less
Submitted 21 June, 2023;
originally announced June 2023.
-
Permutation Invariant Recurrent Neural Networks for Sound Source Tracking Applications
Authors:
David Diaz-Guerra,
Archontis Politis,
Antonio Miguel,
Jose R. Beltran,
Tuomas Virtanen
Abstract:
Many multi-source localization and tracking models based on neural networks use one or several recurrent layers at their final stages to track the movement of the sources. Conventional recurrent neural networks (RNNs), such as the long short-term memories (LSTMs) or the gated recurrent units (GRUs), take a vector as their input and use another vector to store their state. However, this approach re…
▽ More
Many multi-source localization and tracking models based on neural networks use one or several recurrent layers at their final stages to track the movement of the sources. Conventional recurrent neural networks (RNNs), such as the long short-term memories (LSTMs) or the gated recurrent units (GRUs), take a vector as their input and use another vector to store their state. However, this approach results in the information from all the sources being contained in a single ordered vector, which is not optimal for permutation-invariant problems such as multi-source tracking. In this paper, we present a new recurrent architecture that uses unordered sets to represent both its input and its state and that is invariant to the permutations of the input set and equivariant to the permutations of the state set. Hence, the information of every sound source is represented in an individual embedding and the new estimates are assigned to the tracked trajectories regardless of their order.
△ Less
Submitted 14 June, 2023;
originally announced June 2023.
-
Symbolic Music Structure Analysis with Graph Representations and Changepoint Detection Methods
Authors:
Carlos Hernandez-Olivan,
Sonia Rubio Llamas,
Jose R. Beltran
Abstract:
Music Structure Analysis is an open research task in Music Information Retrieval (MIR). In the past, there have been several works that attempt to segment music into the audio and symbolic domains, however, the identification and segmentation of the music structure at different levels is still an open research problem in this area. In this work we propose three methods, two of which are novel grap…
▽ More
Music Structure Analysis is an open research task in Music Information Retrieval (MIR). In the past, there have been several works that attempt to segment music into the audio and symbolic domains, however, the identification and segmentation of the music structure at different levels is still an open research problem in this area. In this work we propose three methods, two of which are novel graph-based algorithms that aim to segment symbolic music by its form or structure: Norm, G-PELT and G-Window. We performed an ablation study with two public datasets that have different forms or structures in order to compare such methods varying their parameter values and comparing the performance against different music styles. We have found that encoding symbolic music with graph representations and computing the novelty of Adjacency Matrices obtained from graphs represent the structure of symbolic music pieces well without the need to extract features from it. We are able to detect the boundaries with an online unsupervised changepoint detection method with a F_1 of 0.5640 for a 1 bar tolerance in one of the public datasets that we used for testing our methods. We also provide the performance results of the algorithms at different levels of structure, high, medium and low, to show how the parameters of the proposed methods have to be adjusted depending on the level. We added the best performing method with its parameters for each structure level to musicaiz, an open source python package, to facilitate the reproducibility and usability of this work. We hope that this methods could be used to improve other MIR tasks such as music generation with structure, music classification or key changes detection.
△ Less
Submitted 24 March, 2023;
originally announced March 2023.
-
Thermodynamic formalism for amenable groups and countable state spaces
Authors:
Elmer R. Beltrán,
Rodrigo Bissacot,
Luísa Borsato,
Raimundo Briceño
Abstract:
Given the full shift over a countable state space on a countable amenable group, we develop its thermodynamic formalism. First, we introduce the concept of pressure and, using tiling techniques, prove its existence and further properties such as an infimum rule. Next, we extend the definitions of different notions of Gibbs measures and prove their existence and equivalence, given some regularity a…
▽ More
Given the full shift over a countable state space on a countable amenable group, we develop its thermodynamic formalism. First, we introduce the concept of pressure and, using tiling techniques, prove its existence and further properties such as an infimum rule. Next, we extend the definitions of different notions of Gibbs measures and prove their existence and equivalence, given some regularity and normalization criteria on the potential. Finally, we provide a family of potentials that non-trivially satisfy the conditions for having this equivalence and a non-empty range of inverse temperatures where uniqueness holds.
△ Less
Submitted 10 February, 2023;
originally announced February 2023.
-
Reinterpretation of searches for long-lived particles from meson decays
Authors:
Rebeca Beltrán,
Giovanna Cottin,
Martin Hirsch,
Arsenii Titov,
Zeren Simon Wang
Abstract:
Many models beyond the Standard Model predict light and feebly interacting particles that are often long-lived. These long-lived particles (LLPs) in many cases can be produced from meson decays. In this work, we propose a simple and quick reinterpretation method for models predicting LLPs produced from meson decays. With the method, we are not required to run Monte-Carlo simulation, implement dete…
▽ More
Many models beyond the Standard Model predict light and feebly interacting particles that are often long-lived. These long-lived particles (LLPs) in many cases can be produced from meson decays. In this work, we propose a simple and quick reinterpretation method for models predicting LLPs produced from meson decays. With the method, we are not required to run Monte-Carlo simulation, implement detector geometries and efficiencies, or apply experimental cuts in an event analysis, as typically done in recasting and reinterpretation works. The main ingredients our method requires are only the theoretical input, allowing for computation of the production and decay rates of the LLPs. There are two conditions for the method to work: firstly, the LLPs in the models considered should be produced from a set of mesons with similar mass and lifetime (or the same meson) and second, the LLPs should, in general, have a lab-frame decay length much larger than the distance between the interaction point and the detector. As an example, we use this method to reinterpret exclusion bounds on heavy neutral leptons (HNLs) in the minimal ``3+1'' scenario, into those for HNLs in the general effective-field-theory framework as well as for axion-like particles. We are able to reproduce existing results, and obtain new bounds via reinterpretation of past experimental results, in particular, from CHARM and Belle.
△ Less
Submitted 12 April, 2023; v1 submitted 6 February, 2023;
originally announced February 2023.
-
A Survey on Artificial Intelligence for Music Generation: Agents, Domains and Perspectives
Authors:
Carlos Hernandez-Olivan,
Javier Hernandez-Olivan,
Jose R. Beltran
Abstract:
Music is one of the Gardner's intelligences in his theory of multiple intelligences. How humans perceive and understand music is still being studied and is crucial to develop artificial intelligence models that imitate such processes. Music generation with Artificial Intelligence is an emerging field that is gaining much attention in the recent years. In this paper, we describe how humans compose…
▽ More
Music is one of the Gardner's intelligences in his theory of multiple intelligences. How humans perceive and understand music is still being studied and is crucial to develop artificial intelligence models that imitate such processes. Music generation with Artificial Intelligence is an emerging field that is gaining much attention in the recent years. In this paper, we describe how humans compose music and how new AI systems could imitate such process by comparing past and recent advances in the field with music composition techniques. To understand how AI models and algorithms generate music and the potential applications that might appear in the future, we explore, analyze and describe the agents that take part of the music generation process: the datasets, models, interfaces, the users and the generated music. We mention possible applications that might benefit from this field and we also propose new trends and future research directions that could be explored in the future.
△ Less
Submitted 3 November, 2022; v1 submitted 25 October, 2022;
originally announced October 2022.
-
Long-lived heavy neutral leptons from mesons in effective field theory
Authors:
Rebeca Beltrán,
Giovanna Cottin,
Juan Carlos Helo,
Martin Hirsch,
Arsenii Titov,
Zeren Simon Wang
Abstract:
In the framework of the low-energy effective field theory of the Standard Model extended with heavy neutral leptons (HNLs), we calculate the production rates of HNLs from meson decays triggered by dimension-six operators. We consider both lepton-number-conserving and lepton-number-violating four-fermion operators involving either a pair of HNLs or a single HNL. Assuming that HNLs are long-lived, w…
▽ More
In the framework of the low-energy effective field theory of the Standard Model extended with heavy neutral leptons (HNLs), we calculate the production rates of HNLs from meson decays triggered by dimension-six operators. We consider both lepton-number-conserving and lepton-number-violating four-fermion operators involving either a pair of HNLs or a single HNL. Assuming that HNLs are long-lived, we perform simulations and investigate the reach of the proposed far detectors at the high-luminosity LHC to (i) active-heavy neutrino mixing and (ii) the Wilson coefficients associated with the effective operators, for HNL masses below the mass of the $B$-meson. We further convert the latter to the associated new-physics scales. Our results show that scales in excess of hundreds of TeV and the active-heavy mixing squared as small as $10^{-15}$ can be probed by these experiments.
△ Less
Submitted 28 November, 2022; v1 submitted 5 October, 2022;
originally announced October 2022.
-
musicaiz: A Python Library for Symbolic Music Generation, Analysis and Visualization
Authors:
Carlos Hernandez-Olivan,
Jose R. Beltran
Abstract:
In this article, we present musicaiz, an object-oriented library for analyzing, generating and evaluating symbolic music. The submodules of the package allow the user to create symbolic music data from scratch, build algorithms to analyze symbolic music, encode MIDI data as tokens to train deep learning sequence models, modify existing music data and evaluate music generation systems. The evaluati…
▽ More
In this article, we present musicaiz, an object-oriented library for analyzing, generating and evaluating symbolic music. The submodules of the package allow the user to create symbolic music data from scratch, build algorithms to analyze symbolic music, encode MIDI data as tokens to train deep learning sequence models, modify existing music data and evaluate music generation systems. The evaluation submodule builds on previous work to objectively measure music generation systems and to be able to reproduce the results of music generation models. The library is publicly available online. We encourage the community to contribute and provide feedback.
△ Less
Submitted 16 September, 2022;
originally announced September 2022.
-
Direction of Arrival Estimation of Sound Sources Using Icosahedral CNNs
Authors:
David Diaz-Guerra,
Antonio Miguel,
Jose R. Beltran
Abstract:
In this paper, we present a new model for Direction of Arrival (DOA) estimation of sound sources based on an Icosahedral Convolutional Neural Network (CNN) applied over SRP-PHAT power maps computed from the signals received by a microphone array. This icosahedral CNN is equivariant to the 60 rotational symmetries of the icosahedron, which represent a good approximation of the continuous space of s…
▽ More
In this paper, we present a new model for Direction of Arrival (DOA) estimation of sound sources based on an Icosahedral Convolutional Neural Network (CNN) applied over SRP-PHAT power maps computed from the signals received by a microphone array. This icosahedral CNN is equivariant to the 60 rotational symmetries of the icosahedron, which represent a good approximation of the continuous space of spherical rotations, and can be implemented using standard 2D convolutional layers, having a lower computational cost than most of the spherical CNNs. In addition, instead of using fully connected layers after the icosahedral convolutions, we propose a new soft-argmax function that can be seen as a differentiable version of the argmax function and allows us to solve the DOA estimation as a regression problem interpreting the output of the convolutional layers as a probability distribution. We prove that using models that fit the equivariances of the problem allows us to outperform other state-of-the-art models with a lower computational cost and more robustness, obtaining root mean square localization errors lower than 10° even in scenarios with a reverberation time $T_{60}$ of 1.5 s.
△ Less
Submitted 6 December, 2022; v1 submitted 31 March, 2022;
originally announced March 2022.
-
Subjective Evaluation of Deep Learning Models for Symbolic Music Composition
Authors:
Carlos Hernandez-Olivan,
Jorge Abadias Puyuelo,
Jose R. Beltran
Abstract:
Deep learning models are typically evaluated to measure and compare their performance on a given task. The metrics that are commonly used to evaluate these models are standard metrics that are used for different tasks. In the field of music composition or generation, the standard metrics used in other fields have no clear meaning in terms of music theory. In this paper, we propose a subjective met…
▽ More
Deep learning models are typically evaluated to measure and compare their performance on a given task. The metrics that are commonly used to evaluate these models are standard metrics that are used for different tasks. In the field of music composition or generation, the standard metrics used in other fields have no clear meaning in terms of music theory. In this paper, we propose a subjective method to evaluate AI-based music composition systems by asking questions related to basic music principles to different levels of users based on their musical experience and knowledge. We use this method to compare state-of-the-art models for music composition with deep learning. We give the results of this evaluation method and we compare the responses of each user level for each evaluated model.
△ Less
Submitted 3 April, 2022; v1 submitted 28 March, 2022;
originally announced March 2022.
-
Existence of the zero temperature limit of equilibrium states on topologically transitive countable Markov shifts
Authors:
Elmer R. Beltrán,
Jorge Littin,
Cesar Maldonado,
Victor Vargas
Abstract:
Consider a topologically transitive countable Markov shift $Σ$ and a summable Markov potential $φ$ with finite Gurevich pressure and $\mathrm{Var}_1(φ) < \infty$. We prove existence of the limit $\lim_{t \to \infty} μ_t$ in the weak$^\star$ topology, where $μ_t$ is the unique equilibrium state associated to the potential $tφ$. Besides that, we present examples where the limit at zero temperature e…
▽ More
Consider a topologically transitive countable Markov shift $Σ$ and a summable Markov potential $φ$ with finite Gurevich pressure and $\mathrm{Var}_1(φ) < \infty$. We prove existence of the limit $\lim_{t \to \infty} μ_t$ in the weak$^\star$ topology, where $μ_t$ is the unique equilibrium state associated to the potential $tφ$. Besides that, we present examples where the limit at zero temperature exists for potentials satisfying more general conditions.
△ Less
Submitted 6 January, 2022;
originally announced January 2022.
-
Long-lived heavy neutral leptons at the LHC: four-fermion single-$N_R$ operators
Authors:
Rebeca Beltrán,
Giovanna Cottin,
Juan Carlos Helo,
Martin Hirsch,
Arsenii Titov,
Zeren Simon Wang
Abstract:
Interest in searches for heavy neutral leptons (HNLs) at the LHC has increased considerably in the past few years. In the minimal scenario, HNLs are produced and decay via their mixing with active neutrinos in the Standard Model (SM) spectrum. However, many SM extensions with HNLs have been discussed in the literature, which sometimes change expectations for LHC sensitivities drastically. In the…
▽ More
Interest in searches for heavy neutral leptons (HNLs) at the LHC has increased considerably in the past few years. In the minimal scenario, HNLs are produced and decay via their mixing with active neutrinos in the Standard Model (SM) spectrum. However, many SM extensions with HNLs have been discussed in the literature, which sometimes change expectations for LHC sensitivities drastically. In the $N_R$SMEFT, one extends the SM effective field theory with operators including SM singlet fermions, which allows to study HNL phenomenology in a "model independent" way. In this paper, we study the sensitivity of ATLAS to HNLs in the $N_R$SMEFT for four-fermion operators with a single HNL. These operators might dominate both production and decay of HNLs, and we find that new physics scales in excess of 20 TeV could be probed at the high-luminosity LHC.
△ Less
Submitted 14 January, 2022; v1 submitted 28 October, 2021;
originally announced October 2021.
-
Submegahertz spectral width photon pair source based on fused silica microspheres
Authors:
Erasto Ortiz Ricardo,
Cesar Bertoni Ocampo,
Mónica Maldonado Terrón,
Arturo Sanchez Zurita,
Roberto Ramirez Alarcón,
Héctor Cruz Ramírez,
Rigoberto Castro Beltran,
Alfred B. U'Ren
Abstract:
High efficiency, sub-MHz bandwidth photon pair generators will enable the field of quantum technology to transition from laboratory demonstrations to transformational applications involving information transfer from photons to atoms. While spontaneous parametric processes are able to achieve high efficiency photon pair generation, the spectral bandwidth tends to be relatively large, as defined by…
▽ More
High efficiency, sub-MHz bandwidth photon pair generators will enable the field of quantum technology to transition from laboratory demonstrations to transformational applications involving information transfer from photons to atoms. While spontaneous parametric processes are able to achieve high efficiency photon pair generation, the spectral bandwidth tends to be relatively large, as defined by phase-matching constraints. To solve this fundamental limitation, we use an ultra-high quality factor (Q) fused silica microsphere resonant cavity to form a photon pair generator. We present the full theory for the SWM process in these devices, fully taking into account all relevant source characteristics in our experiments. The exceptionally narrow (down to kHz-scale) linewidths of these devices in combination with the device size results in a reduction in the bandwidth of the photon pair generation, allowing sub-MHz spectral bandwidth to be achieved. Specifically, using a pump source centered around 1550nm, photon pairs with the signal and idler modes at wavelengths close to 1540nm and 1560nm, respectively, are demonstrated. We herald a single idler-mode photon by detecting the corresponding signal photon, filtered via transmission through a wavelength division multiplexing channel of choice. We demonstrate the extraction of the spectral profile of a single peak in the single-photon frequency comb from a measurement of the signal-idler time of emission distribution. These improvements in device design and experimental methods enabled the narrowest spectral width (366kHz) to date in a heralded single photon source basedon SFWM.
△ Less
Submitted 25 October, 2021;
originally announced October 2021.
-
Music Composition with Deep Learning: A Review
Authors:
Carlos Hernandez-Olivan,
Jose R. Beltran
Abstract:
Generating a complex work of art such as a musical composition requires exhibiting true creativity that depends on a variety of factors that are related to the hierarchy of musical language. Music generation have been faced with Algorithmic methods and recently, with Deep Learning models that are being used in other fields such as Computer Vision. In this paper we want to put into context the exis…
▽ More
Generating a complex work of art such as a musical composition requires exhibiting true creativity that depends on a variety of factors that are related to the hierarchy of musical language. Music generation have been faced with Algorithmic methods and recently, with Deep Learning models that are being used in other fields such as Computer Vision. In this paper we want to put into context the existing relationships between AI-based music composition models and human musical composition and creativity processes. We give an overview of the recent Deep Learning models for music composition and we compare these models to the music composition process from a theoretical point of view. We have tried to answer some of the most relevant open questions for this task by analyzing the ability of current Deep Learning models to generate music with creativity or the similarity between AI and human composition processes, among others.
△ Less
Submitted 7 September, 2021; v1 submitted 27 August, 2021;
originally announced August 2021.
-
Surface Brightness Fluctuations to constrain secondary stellar populations: Revealing very low-metallicity stars in massive galaxies
Authors:
Pablo Rodriguez Beltran,
Alexandre Vazdekis,
Miguel Cervino,
Michael Beasley
Abstract:
The aim of this work is to explore the potential of Surface Brightness Fluctuations (SBF) for studying composite stellar populations (CSP). To do so, we have computed the standard (mean) and SBF spectra with E-MILES stellar population synthesis code. We have created a set of models composed by different mass fractions of two single stellar populations (SSP), as a first approximation of a CSP scena…
▽ More
The aim of this work is to explore the potential of Surface Brightness Fluctuations (SBF) for studying composite stellar populations (CSP). To do so, we have computed the standard (mean) and SBF spectra with E-MILES stellar population synthesis code. We have created a set of models composed by different mass fractions of two single stellar populations (SSP), as a first approximation of a CSP scenario. With these models we present an ensemble of SBF colour-colour diagnostic diagrams that reveal different secondary populations depending on the bands used. For this work we focus on those colours capable of unveiling small fractions of metal-poor components in elliptical galaxies, which are dominated by old metal-rich stellar populations. We fit a set of synthetic models and a selection of nearby elliptical galaxies to our CSP models using both mean and SBF colours. We find that the results are highly improved and return small secondary components when mean and SBF values are applied simultaneously, instead of employing them separately or as a constraint. Finally, we explore the possibility of tracking chemical enrichment histories by including in the analysis a variety of SBF colours. For this purpose we present an example where, with two different SBF colour-colour diagrams, we untangle a small contribution of a young solar population and an old metal-poor component from an old solar principal population. The results we have found are promising, but limited by the available data. We highlight the urgent need for new, better and more consistent SBF observations.
△ Less
Submitted 4 September, 2021; v1 submitted 19 July, 2021;
originally announced July 2021.
-
Timbre Classification of Musical Instruments with a Deep Learning Multi-Head Attention-Based Model
Authors:
Carlos Hernandez-Olivan,
Jose R. Beltran
Abstract:
The aim of this work is to define a model based on deep learning that is able to identify different instrument timbres with as few parameters as possible. For this purpose, we have worked with classical orchestral instruments played with different dynamics, which are part of a few instrument families and which play notes in the same pitch range. It has been possible to assess the ability to classi…
▽ More
The aim of this work is to define a model based on deep learning that is able to identify different instrument timbres with as few parameters as possible. For this purpose, we have worked with classical orchestral instruments played with different dynamics, which are part of a few instrument families and which play notes in the same pitch range. It has been possible to assess the ability to classify instruments by timbre even if the instruments are playing the same note with the same intensity. The network employed uses a multi-head attention mechanism, with 8 heads and a dense network at the output taking as input the log-mel magnitude spectrograms of the sound samples. This network allows the identification of 20 instrument classes of the classical orchestra, achieving an overall F$_1$ value of 0.62. An analysis of the weights of the attention layer has been performed and the confusion matrix of the model is presented, allowing us to assess the ability of the proposed architecture to distinguish timbre and to establish the aspects on which future work should focus.
△ Less
Submitted 13 July, 2021;
originally announced July 2021.
-
Music Boundary Detection using Convolutional Neural Networks: A comparative analysis of combined input features
Authors:
Carlos Hernandez-Olivan,
Jose R. Beltran,
David Diaz-Guerra
Abstract:
The analysis of the structure of musical pieces is a task that remains a challenge for Artificial Intelligence, especially in the field of Deep Learning. It requires prior identification of structural boundaries of the music pieces. This structural boundary analysis has recently been studied with unsupervised methods and \textit{end-to-end} techniques such as Convolutional Neural Networks (CNN) us…
▽ More
The analysis of the structure of musical pieces is a task that remains a challenge for Artificial Intelligence, especially in the field of Deep Learning. It requires prior identification of structural boundaries of the music pieces. This structural boundary analysis has recently been studied with unsupervised methods and \textit{end-to-end} techniques such as Convolutional Neural Networks (CNN) using Mel-Scaled Log-magnitude Spectograms features (MLS), Self-Similarity Matrices (SSM) or Self-Similarity Lag Matrices (SSLM) as inputs and trained with human annotations. Several studies have been published divided into unsupervised and \textit{end-to-end} methods in which pre-processing is done in different ways, using different distance metrics and audio characteristics, so a generalized pre-processing method to compute model inputs is missing. The objective of this work is to establish a general method of pre-processing these inputs by comparing the inputs calculated from different pooling strategies, distance metrics and audio characteristics, also taking into account the computing time to obtain them. We also establish the most effective combination of inputs to be delivered to the CNN in order to establish the most efficient way to extract the limits of the structure of the music pieces. With an adequate combination of input matrices and pooling strategies we obtain a measurement accuracy $F_1$ of 0.411 that outperforms the current one obtained under the same conditions.
△ Less
Submitted 1 December, 2021; v1 submitted 17 August, 2020;
originally announced August 2020.
-
Infinite DLR Measures and Volume-Type Phase Transitions on Countable Markov Shifts
Authors:
Elmer R. Beltrán,
Rodrigo Bissacot,
Eric O. Endo
Abstract:
We consider the natural definition of DLR measure in the setting of $σ$-finite measures on countable Markov shifts. We prove that the set of DLR measures contains the set of conformal measures associated with Walters potentials. In the BIP case, or when the potential normalizes the Ruelle's operator, we prove that the notions of DLR and conformal coincide. On the standard renewal shift, we study t…
▽ More
We consider the natural definition of DLR measure in the setting of $σ$-finite measures on countable Markov shifts. We prove that the set of DLR measures contains the set of conformal measures associated with Walters potentials. In the BIP case, or when the potential normalizes the Ruelle's operator, we prove that the notions of DLR and conformal coincide. On the standard renewal shift, we study the problem of describing the cases when the set of the eigenmeasures jumps from finite to infinite measures when we consider high and low temperatures, respectively. For this particular shift, we prove that there always exist finite DLR measures, and we have an expression to the critical temperature for this volume-type phase transition, which occurs only for potentials with the infinite first variation.
△ Less
Submitted 17 August, 2020; v1 submitted 8 August, 2020;
originally announced August 2020.
-
Robust Sound Source Tracking Using SRP-PHAT and 3D Convolutional Neural Networks
Authors:
David Diaz-Guerra,
Antonio Miguel,
Jose R. Beltran
Abstract:
In this paper, we present a new single sound source DOA estimation and tracking system based on the well-known SRP-PHAT algorithm and a three-dimensional Convolutional Neural Network. It uses SRP-PHAT power maps as input features of a fully convolutional causal architecture that uses 3D convolutional layers to accurately perform the tracking of a sound source even in highly reverberant scenarios w…
▽ More
In this paper, we present a new single sound source DOA estimation and tracking system based on the well-known SRP-PHAT algorithm and a three-dimensional Convolutional Neural Network. It uses SRP-PHAT power maps as input features of a fully convolutional causal architecture that uses 3D convolutional layers to accurately perform the tracking of a sound source even in highly reverberant scenarios where most of the state of the art techniques fail. Unlike previous methods, since we do not use bidirectional recurrent layers and all our convolutional layers are causal in the time dimension, our system is feasible for real-time applications and it provides a new DOA estimation for each new SRP-PHAT map. To train the model, we introduce a new procedure to simulate random trajectories as they are needed during the training, equivalent to an infinite-size dataset with high flexibility to modify its acoustical conditions such as the reverberation time. We use both acoustical simulations on a large range of reverberation times and the actual recordings of the LOCATA dataset to prove the robustness of our system and its good performance even using low-resolution SRP-PHAT maps.
△ Less
Submitted 16 December, 2020; v1 submitted 16 June, 2020;
originally announced June 2020.
-
gpuRIR: A Python Library for Room Impulse Response Simulation with GPU Acceleration
Authors:
David Diaz-Guerra,
Antonio Miguel,
Jose R. Beltran
Abstract:
The Image Source Method (ISM) is one of the most employed techniques to calculate acoustic Room Impulse Responses (RIRs), however, its computational complexity grows fast with the reverberation time of the room and its computation time can be prohibitive for some applications where a huge number of RIRs are needed. In this paper, we present a new implementation that dramatically improves the compu…
▽ More
The Image Source Method (ISM) is one of the most employed techniques to calculate acoustic Room Impulse Responses (RIRs), however, its computational complexity grows fast with the reverberation time of the room and its computation time can be prohibitive for some applications where a huge number of RIRs are needed. In this paper, we present a new implementation that dramatically improves the computation speed of the ISM by using Graphic Processing Units (GPUs) to parallelize both the simulation of multiple RIRs and the computation of the images inside each RIR. Additional speedups were achieved by exploiting the mixed precision capabilities of the newer GPUs and by using lookup tables. We provide a Python library under GNU license that can be easily used without any knowledge about GPU programming and we show that it is about 100 times faster than other state of the art CPU libraries. It may become a powerful tool for many applications that need to perform a large number of acoustic simulations, such as training machine learning systems for audio signal processing, or for real-time room acoustics simulations for immersive multimedia systems, such as augmented or virtual reality.
△ Less
Submitted 9 October, 2020; v1 submitted 26 October, 2018;
originally announced October 2018.
-
Stability of As$_n$ [n=4, 8, 20, 28, 32, 36, 60] Cage Structures
Authors:
Tunna Baruah,
Mark R. Pederson,
Rajendra R. Zope,
M. R. Beltran
Abstract:
We present all-electron density functional study of the geometry, electronic structure, vibrational modes, polarizabilities as well as the infrared and Raman spectra of fullerene-like arsenic cages. The stability of As$_n$ cages for sizes 4, 8, 20, 28, 32, 36, and 60 wherein each As atom is three-fold coordinated is examined. We find that all the cages studied are vibrationally stable and while…
▽ More
We present all-electron density functional study of the geometry, electronic structure, vibrational modes, polarizabilities as well as the infrared and Raman spectra of fullerene-like arsenic cages. The stability of As$_n$ cages for sizes 4, 8, 20, 28, 32, 36, and 60 wherein each As atom is three-fold coordinated is examined. We find that all the cages studied are vibrationally stable and while all the clusters are energetically stable with respect to isolated arsenic atoms, only As$_{20}$ is energetically stable against dissociation into As$_4$. We suggest that the Raman spectra might be a means for observing the As$_{20}$ molecule in gas phase.
△ Less
Submitted 19 February, 2004;
originally announced February 2004.
-
Melting of sodium clusters
Authors:
Juan A. Reyes-Nava,
Ignacio L. Garzon,
Marcela R. Beltran,
Karo Michaelian
Abstract:
Thermal stability properties and the melting-like transition of Na_n, n=13-147, clusters are studied through microcanonical molecular dynamics simulations. The metallic bonding in the sodium clusters is mimicked by a many-body Gupta potential based on the second moment approximation of a tight-binding Hamiltonian. The characteristics of the solid-to-liquid transition in the sodium clusters are a…
▽ More
Thermal stability properties and the melting-like transition of Na_n, n=13-147, clusters are studied through microcanonical molecular dynamics simulations. The metallic bonding in the sodium clusters is mimicked by a many-body Gupta potential based on the second moment approximation of a tight-binding Hamiltonian. The characteristics of the solid-to-liquid transition in the sodium clusters are analyzed by calculating physical quantities like caloric curves, heat capacities, and root-mean-square bond length fluctuations using simulation times of several nanoseconds. Distinct melting mechanisms are obtained for the sodium clusters in the size range investigated. The calculated melting temperatures show an irregular variation with the cluster size, in qualitative agreement with recent experimental results. However, the calculated melting point for the Na_55 cluster is about 40 % lower than the experimental value.
△ Less
Submitted 17 July, 2002;
originally announced July 2002.
-
Chirality in Bare and Passivated Gold Nanoclusters
Authors:
I. L. Garzon,
J. A. Reyes-Nava,
J. I. Rodriguea-Hernandez,
I. Sigal,
M. R. Beltran,
K. Michaelian
Abstract:
Chiral structures have been found as the lowest-energy isomers of bare (Au$_{28}$ and Au$_{55}) and thiol-passivated (Au$_{28}(SCH$_{3})$_{16}$ and Au$_{38}(SCH_{3})_{24}) gold nanoclusters. The degree of chirality existing in the chiral clusters was calculated using the Hausdorff chirality measure. We found that the index of chirality is higher in the passivated clusters and decreases with the…
▽ More
Chiral structures have been found as the lowest-energy isomers of bare (Au$_{28}$ and Au$_{55}) and thiol-passivated (Au$_{28}(SCH$_{3})$_{16}$ and Au$_{38}(SCH_{3})_{24}) gold nanoclusters. The degree of chirality existing in the chiral clusters was calculated using the Hausdorff chirality measure. We found that the index of chirality is higher in the passivated clusters and decreases with the cluster size. These results are consistent with the observed chiroptical activity recently reported for glutahione-passivated gold nanoclusters, and provide theoretical support for the existence of chirality in these novel compounds.
△ Less
Submitted 26 March, 2002;
originally announced March 2002.
-
Gause's exclusion principle revisited: artificial modified species and competition
Authors:
J. C. Flores,
R. Beltran
Abstract:
Gause's principle of competition between two species is studied when one of them is sterile. We study the condition for total extinction in the niche, namely, when the sterile population exterminates the native one by an optimal use of resources. A mathematical Lotka-Volterra non linear model of interaction between a native and sterile species is proposed. The condition for total extinction is r…
▽ More
Gause's principle of competition between two species is studied when one of them is sterile. We study the condition for total extinction in the niche, namely, when the sterile population exterminates the native one by an optimal use of resources. A mathematical Lotka-Volterra non linear model of interaction between a native and sterile species is proposed. The condition for total extinction is related to the initial number $M_{o}$ of sterile individuals released in the niche. In fact, the existence of a critical sterile-population value $M_{c}$ is conjectured from numerical analysis and an analytical estimation is found. When spatial diffusion (migration) is considered a critical size territory is found and, for small territory, total extinction exist in any case. This work is motived by the extermination agriculture problem of fruit flies in our region.
△ Less
Submitted 26 July, 2000;
originally announced July 2000.
