Search

During training, weight matrices in machine learning architectures are updated using stochastic gradient descent or variations thereof. In this contribution we employ concepts of random matrix theory to analyse the resulting stochastic matrix dynamics. We first demonstrate that the dynamics can generically be described using Dyson Brownian motion, leading to e.g. eigenvalue repulsion. The level of stochasticity is shown to depend on the ratio of the learning rate and the mini-batch size, explaining the empirically observed linear scaling rule. We verify this linear scaling in the restricted Boltzmann machine. Subsequently we study weight matrix dynamics in transformers (a nano-GPT), following the evolution from a Marchenko-Pastur distribution for eigenvalues at initialisation to a combination with additional structure at the end of learning., Comment: 7 pages. Contribution accepted in the NeurIPS 2024 workshop "Machine Learning and the Physical Sciences"

First-order phase transitions in the early universe have rich phenomenological implications, such as the production of a potentially detectable signal of stochastic relic background gravitational waves. The hypothesis that new, strongly coupled dynamics, hiding in a new dark sector, could be detected in this way, via the telltale signs of its confinement/deconfinement phase transition, provides a fascinating opportunity for interdisciplinary synergy between lattice field theory and astro-particle physics. But its viability relies on completing the challenging task of providing accurate theoretical predictions for the parameters characterising the strongly coupled theory. Density of states methods, and in particular the linear logarithmic relaxation (LLR) method, can be used to address the intrinsic numerical difficulties that arise due the meta-stable dynamics in the vicinity of the critical point. For example, it allows one to obtain accurate determinations of thermodynamic observables that are otherwise inaccessible, such as the free energy. In this contribution, we present an update on results of the analysis of the finite temperature deconfinement phase transition in a pure gauge theory with a symplectic gauge group, $Sp(4)$, by using the LLR method. We present a first analysis of the properties of the transition in the thermodynamic limit, and provide a road map for future work, including a brief preliminary discussion that will inform future publications., Comment: 13 pages, 3 figures. Contribution to Proceedings of the 41st International Symposium on Lattice Field Theory (Lattice 2024), July 28th - August 3rd, 2024, University of Liverpool, UK

In this paper, we apply our minimax theory ([4], [5], [6]) with the one developed by A. Moameni in [2] to formalize a general scheme giving the multiplicity of critical points. Here is a sample of application of the scheme to a critical elliptic problem: Let $\Omega\subset {\bf R}^n$ ($n\geq 3$) be a smooth bounded domain and let $10$, there exists $\lambda^*>0$ with the following property: for every $\lambda\in ]0,\lambda^*[$, $\mu\in ]-\lambda^*,\lambda^*[$, and for every convex dense set $S\subset H^{-1}(\Omega)$, there exists $\tilde\varphi\in S$, with $\|\tilde\varphi\|_{H^{-1}(\Omega)}

Industrial applications often exhibit multiple in-control patterns due to varying operating conditions, which makes a single functional linear model (FLM) inadequate to capture the complexity of the true relationship between a functional quality characteristic and covariates, which gives rise to the multimode profile monitoring problem. This issue is clearly illustrated in the resistance spot welding (RSW) process in the automotive industry, where different operating conditions lead to multiple in-control states. In these states, factors such as electrode tip wear and dressing may influence the functional quality characteristic differently, resulting in distinct FLMs across subpopulations. To address this problem, this article introduces the functional mixture regression control chart (FMRCC) to monitor functional quality characteristics with multiple in-control patterns and covariate information, modeled using a mixture of FLMs. A monitoring strategy based on the likelihood ratio test is proposed to monitor any deviation from the estimated in-control heterogeneous population. An extensive Monte Carlo simulation study is performed to compare the FMRCC with competing monitoring schemes that have already appeared in the literature, and a case study in the monitoring of an RSW process in the automotive industry, which motivated this research, illustrates its practical applicability.

The family of SU(2) theories with matter transforming in the adjoint representation has attracted interest from many angles. The two-flavour theory, known as Minimal Walking Technicolor, has a body of evidence pointing to it being in the conformal window with anomalous dimension $\gamma_{*}\approx0.3$. Perturbative calculations would suggest that the one-flavour theory should be confining and chirally broken; however, lattice studies of the theory have been inconclusive. In this contribution we present a first look at efforts towards the computation of the beta function of these theories using the gradient flow methodology. Following an exploration of the phase diagram of the two theories with Wilson fermions and additional Pauli-Villars fields, we tune the bare fermion mass to near the chiral limit, and subsequently generate ensembles at five lattice volumes and a range of lattice spacings., Comment: 9 pages, 7 figures. Contribution to Proceedings of the 41st International Symposium on Lattice Field Theory (Lattice 2024), July 28th - August 3rd, 2024, University of Liverpool, UK

We measure the masses of the pseudoscalar flavour-singlet meson states in the $Sp(4)$ gauge theory coupled to two Dirac fermions transforming in the fundamental representation and three Dirac fermions in the antisymmetric representation. This theory provides a compelling ultraviolet completion for the minimal composite Higgs model implementing also partial compositeness for the top quark. The spectrum contains two, comparatively light, pseudoscalar flavour-singlet states, which mix with one another. One of them is a Nambu-Goldstone boson (in the massless limit), whereas the other receives a mass from the $U(1)_A$ axial anomaly. We demonstrate how to measure the mixing between these two states. For moderately heavy fermion masses, we find that the two wave functions are dominated by one of the fermion representations, mixing effects being small., Comment: 12 pages, 2 figures, 2 tables, Proceedings of the 41st International Symposium on Lattice Field Theory (Lattice 2024), July 28th - August 3rd, 2024, University of Liverpool, UK

Spectral densities encode non-perturbative information crucial in computing physical observables in strongly coupled field theories. Using lattice gauge theory data, we perform a systematic study to demonstrate the potential of recent technological advances in the reconstruction of spectral densities. We develop, maintain and make publicly available dedicated analysis code that can be used for broad classes of lattice theories. As a test case, we analyse the Sp(4) gauge theory coupled to an admixture of fermions transforming in the fundamental and two-index antisymmetric representations. We measure the masses of mesons in energy-smeared spectral densities, after optimising the smearing parameters for available lattice ensembles. We present a summary of the mesons mass spectrum in all the twelve (flavored) channels available, including also several excited states., Comment: 9 pages, 5 figures, contribution for the 41th International Symposium on Lattice Field Theory (Lattice 2024), 28 July - 3 August 2024, Liverpool, UK, presenting the results of the paper: arXiv:2405.01388

In this paper, given a topological space $X$, an interval $I\subseteq {\bf R}$ and five continuous functions $\varphi, \psi, \omega :X\to {\bf R}$, $\alpha, \beta:I\to {\bf R}$, we are interested in the infimum of the function $\Phi:X\to ]-\infty,+\infty]$ defined by $$\Phi(x)=\sup_{\lambda\in I}(\alpha(\lambda)\varphi(x)+\beta(\lambda)\psi(x))+\omega(x)\ .$$ Using a recent minimax theorem ([5]), we build a general scheme which provides the exact value of $\inf_X\Phi$ for a large class of functions $\Phi$. When additional compactness conditions are satisfied, our scheme provides also the existence of (explicitly detected) functions $\gamma, \eta:X\to {\bf R}$ such that, for some $\tilde x\in X$, one has $$\gamma(\tilde x)\varphi(\tilde x)+\eta(\tilde x)\psi(\tilde x)+\omega(\tilde x)=\inf_{x\in X}(\gamma(\tilde x)\varphi(x)+\eta(\tilde x)\psi(x)+\omega(x))\ .$$

The problem of control based on vision measurements (bearings) has been amply studied in the literature; however, the problem of addressing the limits of the field of view of physical sensors has received relatively less attention (especially for agents with non-trivial dynamics). The technical challenge is that, as in most vision-based control approaches, a standard approach to the problem requires knowing the distance between cameras and observed features in the scene, which is not directly available. Instead, we present a solution based on a Control Barrier Function (CBF) approach that uses a splitting of the original differential constraint to effectively remove the dependence on the unknown measurement error. Compared to the current literature, our approach gives strong robustness guarantees against bounded distance estimation errors. We showcase the proposed solution with the numerical simulations of a double integrator and a quadrotor tracking a trajectory while keeping the corners of a rectangular gate in the camera field of view., Comment: 8 pages, conference paper

We study the finite-temperature behaviour of the $Sp(4)$ Yang-Mills lattice theory in four dimensions, by applying the Logarithmic Linear Relaxation (LLR) algorithm. We demonstrate the presence of coexisting (metastable) phases, when the system is in the proximity of the transition. We measure observables such as the free energy, the expectation value of the plaquette operator and of the Polyakov loop, as well as the specific heat, and the Binder cumulant. We use these results to obtain a high-precision measurement of the critical coupling at the confinement-deconfinement transition, and assess its systematic uncertainty, for one value of the lattice extent in the time direction. Furthermore, we perform an extensive study of the finite-volume behaviour of the lattice system, by repeating the measurements for fixed lattice time extent, while increasing the spatial size of the lattice. We hence characterise the first-order transition on the lattice, and present the first results in the literature on this theory for the infinite volume extrapolation of lattice quantities related to latent heat and interface tension. Gauge theories with $Sp(4)$ group have been proposed as new dark sectors to provide a fundamental origin for the current phenomenological evidence of dark matter. A phase transition at high temperature, in such a new dark sector, occurring in the early universe, might have left a relic stochastic background of gravitational waves. Our results represent a milestone toward establishing whether such a new physics signal is detectable in future experiments, as they enter the calculation of the parameters, $\alpha$ and $\beta$, controlling the power spectrum of gravitational waves. We also outline the process needed in the continuum extrapolation of our measurements, and test its feasibility on one additional choice of temporal extent of the lattice., Comment: 27 pages, 18 figures

In this paper, a statistical analysis of the performance of a low insulin index, alkaline and functional diet developed by ANTUR evaluated with machine learning techniques is reported. The sample of patients was checked on a regular basis with a BioImpedenziometric (BIA) analysis. The BIA gives about 40 parameters output describing the health status of the patient. The sample of 1626 patients was grouped in clusters with similar characteristics by a neural network algorithm. A study of the behaviour of the BIA parameters evolution over time with respect to the first visit was performed. More than the 75% of the patient that followed the ANTUR diet showed an improvement of the health status.

The Lattice Boltzmann Method (LBM) is a computational technique of Computational Fluid Dynamics (CFD) that has gained popularity due to its high parallelism and ability to handle complex geometries with minimal effort. Although LBM frameworks are increasingly important in various industries and research fields, their complexity makes them difficult to modify and can lead to suboptimal performance. This paper presents miniLB, the first, to the best of our knowledge, SYCL-based LBM mini-app.miniLB addresses the need for a performance-portable LBM proxy app capable of abstracting complex fluid dynamics simulations across heterogeneous computing systems. We analyze SYCL semantics for performance portability and evaluate miniLB on multiple GPU architectures using various SYCL implementations. Our results, compared against a manually-tuned FORTRAN version, demonstrate effectiveness of miniLB in assessing LBM performance across diverse hardware, offering valuable insights for optimizing large-scale LBM frameworks in modern computing environments., Comment: 12 pages, conference

We provide an extended lattice study of the SU(2) gauge theory coupled to one Dirac fermion flavour ($N_{\mathrm{f}} =1$) transforming in the adjoint representation as the continuum limit is approached. This investigation is supplemented by numerical results obtained for the SU(2) gauge theory with two Dirac fermion flavours ($N_{\mathrm{f}} =2$) transforming in the adjoint representation, for which we perform numerical investigations at a single lattice spacing value, which is analysed together with earlier calculations. The purpose of our study is to advance the characterisation of the infrared properties of both theories, which previous investigations have concluded to be in the conformal window. For both, we determine the mass spectrum and the anomalous dimension of the fermion condensate using finite-size hyperscaling of the spectrum, mode number analysis of the Dirac operator (for which we improve on our previous proposal) and the ratio of masses of the lightest spin-2 particle over the lightest scalar. All methods provide a consistent picture, with the anomalous dimension of the condensate $\gamma_*$ decreasing significantly as one approaches the continuum limit for the $N_{\mathrm{f}} = 1$ theory towards a value consistent with $\gamma_* = 0.174(6)$, while for $N_{\mathrm{f}} = 2$ the anomalous dimension decreases more slowly with $\beta$. A chiral perturbation theory analysis show that the infrared behaviour of both theories is incompatible with the breaking of chiral symmetry., Comment: 24 pages, 22 figures

Despite advancements in methodologies, immunohistochemistry (IHC) remains the most utilized ancillary test for histopathologic and companion diagnostics in targeted therapies. However, objective IHC assessment poses challenges. Artificial intelligence (AI) has emerged as a potential solution, yet its development requires extensive training for each cancer and IHC type, limiting versatility. We developed a Universal IHC (UIHC) analyzer, an AI model for interpreting IHC images regardless of tumor or IHC types, using training datasets from various cancers stained for PD-L1 and/or HER2. This multi-cohort trained model outperforms conventional single-cohort models in interpreting unseen IHCs (Kappa score 0.578 vs. up to 0.509) and consistently shows superior performance across different positive staining cutoff values. Qualitative analysis reveals that UIHC effectively clusters patches based on expression levels. The UIHC model also quantitatively assesses c-MET expression with MET mutations, representing a significant advancement in AI application in the era of personalized medicine and accumulating novel biomarkers.

We demonstrate that the update of weight matrices in learning algorithms can be described in the framework of Dyson Brownian motion, thereby inheriting many features of random matrix theory. We relate the level of stochasticity to the ratio of the learning rate and the mini-batch size, providing more robust evidence to a previously conjectured scaling relationship. We discuss universal and non-universal features in the resulting Coulomb gas distribution and identify the Wigner surmise and Wigner semicircle explicitly in a teacher-student model and in the (near-)solvable case of the Gaussian restricted Boltzmann machine., Comment: 17 pages, 16 figures

Despite their impact on the society, deep neural networks are often regarded as black-box models due to their intricate structures and the absence of explanations for their decisions. This opacity poses a significant challenge to AI systems wider adoption and trustworthiness. This thesis addresses this issue by contributing to the field of eXplainable AI, focusing on enhancing the interpretability of deep neural networks. The core contributions lie in introducing novel techniques aimed at making these networks more interpretable by leveraging an analysis of their inner workings. Specifically, the contributions are threefold. Firstly, the thesis introduces designs for self-explanatory deep neural networks, such as the integration of external memory for interpretability purposes and the usage of prototype and constraint-based layers across several domains. Secondly, this research delves into novel investigations on neurons within trained deep neural networks, shedding light on overlooked phenomena related to their activation values. Lastly, the thesis conducts an analysis of the application of explanatory techniques in the field of visual analytics, exploring the maturity of their adoption and the potential of these systems to convey explanations to users effectively., Comment: PhD Thesis

The primary goal of structural health monitoring is to detect damage at its onset before it reaches a critical level. The in-depth investigation in the present work addresses deep learning applied to data-driven damage detection in nonlinear dynamic systems. In particular, autoencoders (AEs) and generative adversarial networks (GANs) are implemented leveraging on 1D convolutional neural networks. The onset of damage is detected in the investigated nonlinear dynamic systems by exciting random vibrations of varying intensity, without prior knowledge of the system or the excitation and in unsupervised manner. The comprehensive numerical study is conducted on dynamic systems exhibiting different types of nonlinear behavior. An experimental application related to a magneto-elastic nonlinear system is also presented to corroborate the conclusions., Comment: 34 pages, 17 figures, 4 tables

ModSecurity is widely recognized as the standard open-source Web Application Firewall (WAF), maintained by the OWASP Foundation. It detects malicious requests by matching them against the Core Rule Set (CRS), identifying well-known attack patterns. Each rule is manually assigned a weight based on the severity of the corresponding attack, and a request is blocked if the sum of the weights of matched rules exceeds a given threshold. However, we argue that this strategy is largely ineffective against web attacks, as detection is only based on heuristics and not customized on the application to protect. In this work, we overcome this issue by proposing a machine-learning model that uses the CRS rules as input features. Through training, ModSec-Learn is able to tune the contribution of each CRS rule to predictions, thus adapting the severity level to the web applications to protect. Our experiments show that ModSec-Learn achieves a significantly better trade-off between detection and false positive rates. Finally, we analyze how sparse regularization can reduce the number of rules that are relevant at inference time, by discarding more than 30% of the CRS rules. We release our open-source code and the dataset at https://github.com/pralab/modsec-learn and https://github.com/pralab/http-traffic-dataset, respectively., Comment: arXiv admin note: text overlap with arXiv:2308.04964

We provide the first extensive, numerical study of the non-trivial problem of mixing between flavor-singlet composite states emerging in strongly coupled lattice field theories with matter field content consisting of fermions transforming in different representations of the gauge group. The theory of interest is the minimal candidate for a composite Higgs model that also accommodates a mechanism for top partial compositeness: the $Sp(4)$ gauge theory coupled to two (Dirac) fermions transforming as the fundamental and three as the two-index antisymmetric representation of the gauge group, respectively. We apply an admixture of APE and Wuppertal smearings, as well as the generalized eigenvalue problem approach, to two-point functions involving flavor-singlet mesons, for ensembles having time extent longer than the space extent. We demonstrate that, in the region of lattice parameter space accessible to this study, both masses and mixing angles can be measured effectively, despite the presence of (numerically noisy) contributions from disconnected diagrams., Comment: 18 pages, 3 figures, 5 tables; v2: minor changes, added appendix A, version accepted for publication in PRD

Let $H$ be a real Hilbert space and $\Phi:H\to H$ be a $C^1$ operator with Lipschitzian derivative and closed range. We prove that $\Phi^{-1}(0)\neq \emptyset$ if and only if, for each $\epsilon>0$, there exist a convex set $X\subset H$ and a convex function $\psi:X\to {\bf R}$ such that $\sup_{x\in X}(\|x\|^2+\psi(x))-\inf_{x\in X}\|x\|^2+\psi(x))<\epsilon$ and $0\in \overline{conv}(\Phi(X))$., Comment: Accepted in Canad. Math. Bull

Spectral densities encode non-perturbative information that enters the calculation of a plethora of physical observables in strongly coupled field theories. Phenomenological applications encompass aspects of standard-model hadronic physics, observable at current colliders, as well as correlation functions characterizing new physics proposals, testable in future experiments. By making use of numerical data produced in a Sp(4) lattice gauge theory with matter transforming in an admixture of fundamental and 2-index antisymmetric representations of the gauge group, we perform a systematic study to demonstrate the effectiveness of recent technological progress in the reconstruction of spectral densities. To this purpose, we write and test new software packages that use energy-smeared spectral densities to analyze the mass spectrum of mesons. We assess the effectiveness of different smearing kernels and optimize the smearing parameters to the characteristics of available lattice ensembles. We generate new ensembles for the theory in consideration, with lattices that have a longer extent in the time direction with respect to the spatial ones. We run our tests on these ensembles, obtaining new results about the spectrum of light mesons and their excitations. We make available our algorithm and software for the extraction of spectral densities, that can be applied to theories with other gauge groups, including the theory of strong interactions (QCD) governing hadronic physics in the standard model., Comment: Version approved for publication. 45 pages, 20 figures

We compare three popular neural architecture search strategies on chain-structured search spaces: Bayesian optimization, the hyperband method, and reinforcement learning in the context of financial time series forecasting., Comment: 17 pages, 3 figures

In $4$-dimensional pure compact $U(1)$ lattice gauge theory, we analyse topological aspects of the dynamics of monopoles across the deconfinement phase transition. We do this using tools from Topological Data Analysis (TDA). We demonstrate that observables constructed from the zeroth and first homology groups of monopole current networks may be used to quantitatively and robustly locate the critical inverse coupling $\beta_{c}$ through finite-size scaling. Our method provides a mathematically robust framework for the characterisation of topological invariants related to monopole currents, putting on firmer ground earlier investigations. Moreover, our approach can be generalised to the study of Abelian monopoles in non-Abelian gauge theories., Comment: 26 pages, 8 figures, 6 tables, resubmission to SciPost; v4: clarifications added; typos corrected; analysis, results and conclusions unchanged

In many modern industrial scenarios, the measurements of the quality characteristics of interest are often required to be represented as functional data or profiles. This motivates the growing interest in extending traditional univariate statistical process monitoring (SPM) schemes to the functional data setting. This article proposes a new SPM scheme, which is referred to as adaptive multivariate functional EWMA (AMFEWMA), to extend the well-known exponentially weighted moving average (EWMA) control chart from the univariate scalar to the multivariate functional setting. The favorable performance of the AMFEWMA control chart over existing methods is assessed via an extensive Monte Carlo simulation. Its practical applicability is demonstrated through a case study in the monitoring of the quality of a resistance spot welding process in the automotive industry through the online observations of dynamic resistance curves, which are associated with multiple spot welds on the same car body and recognized as the full technological signature of the process.

Experimental proposals for testing quantum gravity-induced entanglement of masses (QGEM) typically involve two interacting masses which are each in a spatial superposition state. Here, we propose instead a QGEM experiment with two particles which are each in a superposition of rotational states, this amounts to a superposition of mass through mass-energy equivalence. In sharp contrast to the typical protocols studied, our proposal is genuinely relativistic. It does not consider a quantum positional degree of freedom but relies on the fact that rotational energy gravitates: the effect we consider disappears in the limit where the speed of light c approaches infinity. Furthermore, this approach would test a feature unique to gravity since it amounts to sourcing a spacetime in superposition due to a superposition of 'charge'., Comment: 5 pages, 3 figures, 3 pages supplemental material. Updated manuscript, accepted by Phys Rev D

An influential theorem by Satosi Watanabe convinced many that there can be no physical probabilistic theory with both non-trivial forward and backward transition probabilities. We show that this conclusion does not follow from the theorem. We point out the flaw in the argument, and we showcase examples of theories with well-defined backward and forward transition probabilities., Comment: 10 single column pages, no figures, version accepted for publication, improved thanks to peer review

In strongly coupled gauge theories that serve as completions of composite Higgs models, the fermionic bound states formed by fermions (hyperquarks) transforming in different representations, called chimera baryons, could serve as top partners, by embedding of the Standard Model appropriately. We report our results on the spectrum of chimera baryons in the Sp(4) gauge theory with hyperquarks transforming in fundamental and two-index antisymmetric representations. For this study, we adopt the quenched approximation. We investigate the mass hierarchy between the lightest chimera baryons with different quantum numbers, as a function of the lattice parameters. Inspired by baryon chiral effective field theory, and the Akaike Information Criterion, we perform a first extrapolation to the continuum and massless-hyperquark limit., Comment: 8 pages, 2 figures, contribution to the proceedings of the 40th International Symposium on Lattice Field Theory (LATTICE2023), July 31st - August 4th, 2023, Fermi National Accelerator Laboratory, Batavia, Illinois, USA