Search

We present the CARLA corner case simulation (3CSim) for evaluating autonomous driving (AD) systems within the CARLA simulator. This framework is designed to address the limitations of traditional AD model training by focusing on non-standard, rare, and cognitively challenging scenarios. These corner cases are crucial for ensuring vehicle safety and reliability, as they test advanced control capabilities under unusual conditions. Our approach introduces a taxonomy of corner cases categorized into state anomalies, behavior anomalies, and evidence-based anomalies. We implement 32 unique corner cases with adjustable parameters, including 9 predefined weather conditions, timing, and traffic density. The framework enables repeatable and modifiable scenario evaluations, facilitating the creation of a comprehensive dataset for further analysis.

We developed a modular silicon photomultiplier camera to detect Earth-skimming PeV to EeV tau neutrinos with the imaging atmospheric Cherenkov technique. We built two cameras, a 256-pixel camera with S14161-6050HS SiPMs for the Trinity Demonstrator located on Frisco Peak, Utah, and a 512-pixel camera with S14521-6050AN SiPMs for the EUSO-SPB2 Cherenkov Telescope. The front-end electronics are based on the eMUSIC ASIC, and the camera signals are sampled and digitized with the 100MS/s and 12-bit AGET system. Both cameras are liquid-cooled. We detail the camera concept and the results from characterizing the SiPMs, bench testing, and calibrating the two cameras., Comment: Submitted to Nuclear Instruments and Methods in Physics Research A

Transthoracic Echocardiography (TTE) is a fundamental, non-invasive diagnostic tool in cardiovascular medicine, enabling detailed visualization of cardiac structures crucial for diagnosing various heart conditions. Despite its widespread use, TTE ultrasound imaging faces inherent limitations, notably the trade-off between field of view (FoV) and resolution. This paper introduces a novel application of conditional Generative Adversarial Networks (cGANs), specifically designed to extend the FoV in TTE ultrasound imaging while maintaining high resolution. Our proposed cGAN architecture, termed echoGAN, demonstrates the capability to generate realistic anatomical structures through outpainting, effectively broadening the viewable area in medical imaging. This advancement has the potential to enhance both automatic and manual ultrasound navigation, offering a more comprehensive view that could significantly reduce the learning curve associated with ultrasound imaging and aid in more accurate diagnoses. The results confirm that echoGAN reliably reproduce detailed cardiac features, thereby promising a significant step forward in the field of non-invasive cardiac naviagation and diagnostics.

The Pauli matrices are 2-by-2 matrices that are very useful in quantum computing. They can be used as elementary gates in quantum circuits but also to decompose any matrix of $\mathbb{C}^{2^n \times 2^n}$ as a linear combination of tensor products of the Pauli matrices. However, the computational cost of this decomposition is potentially very expensive since it can be exponential in $n$. In this paper, we propose an algorithm with a parallel implementation that optimizes this decomposition using a tree approach to avoid redundancy in the computation while using a limited memory footprint. We also explain how some particular matrix structures can be exploited to reduce the number of operations. We provide numerical experiments to evaluate the sequential and parallel performance of our decomposition algorithm and we illustrate how this algorithm can be applied to encode matrices in a quantum memory., Comment: 11 pages, 7 figures, submitted to ISC 2024

The metaverse is a virtual space that combines physical and digital elements, creating immersive and connected digital worlds. For autonomous mobility, it enables new possibilities with edge computing and digital twins (DTs) that offer virtual prototyping, prediction, and more. DTs can be created with 3D scene reconstruction methods that capture the real world's geometry, appearance, and dynamics. However, sending data for real-time DT updates in the metaverse, such as camera images and videos from connected autonomous vehicles (CAVs) to edge servers, can increase network congestion, costs, and latency, affecting metaverse services. Herein, a new method is proposed based on distributed radiance fields (RFs), multi-access edge computing (MEC) network for video compression and metaverse DT updates. RF-based encoder and decoder are used to create and restore representations of camera images. The method is evaluated on a dataset of camera images from the CARLA simulator. Data savings of up to 80% were achieved for H.264 I-frame - P-frame pairs by using RFs instead of I-frames, while maintaining high peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM) qualitative metrics for the reconstructed images. Possible uses and challenges for the metaverse and autonomous mobility are also discussed., Comment: 6 pages, 6 figures, conference

We design an algorithm for computing the $L$-series associated to an Anderson $t$-motives, exhibiting quasilinear complexity with respect to the target precision. Based on experiments, we conjecture that the order of vanishing at $T=1$ of the $v$-adic $L$-series of a given Anderson $t$-motive with good reduction does not depend on the finite place $v$.

The Extreme Universe Space Observatory on a Super Pressure Balloon 1 (EUSO-SPB1) was launched in 2017 April from Wanaka, New Zealand. The plan of this mission of opportunity on a NASA super pressure balloon test flight was to circle the southern hemisphere. The primary scientific goal was to make the first observations of ultra-high-energy cosmic-ray extensive air showers (EASs) by looking down on the atmosphere with an ultraviolet (UV) fluorescence telescope from suborbital altitude (33~km). After 12~days and 4~hours aloft, the flight was terminated prematurely in the Pacific Ocean. Before the flight, the instrument was tested extensively in the West Desert of Utah, USA, with UV point sources and lasers. The test results indicated that the instrument had sensitivity to EASs of approximately 3 EeV. Simulations of the telescope system, telescope on time, and realized flight trajectory predicted an observation of about 1 event assuming clear sky conditions. The effects of high clouds were estimated to reduce this value by approximately a factor of 2. A manual search and a machine-learning-based search did not find any EAS signals in these data. Here we review the EUSO-SPB1 instrument and flight and the EAS search., Comment: 18 pages, 19 figures

This is a collection of papers presented by the JEM-EUSO Collaboration at the 38th International Cosmic Ray Conference (Nagoya, Japan, July 26-August 3, 2023)

Over the last decade, conflicting values of the hypertriton ${}_{\Lambda}^3\mathrm{H}$ lifetime $\tau({}_{\Lambda}^3\mathrm{H})$ were extracted from relativistic heavy-ion (RHI) collision experiments, ranging from values compatible with the free-$\Lambda$ lifetime $\tau_\Lambda$-as expected naively for a very weakly bound $\Lambda$ in ${}_{\Lambda}^3\mathrm{H}$-to lifetimes as short as $\tau({}_{\Lambda}^3\mathrm{H}) \approx (0.4-0.7)\,\tau_\Lambda$. In a recent work [Phys. Lett. B 811, 135916 (2020)] we studied this ${}_{\Lambda}^3\mathrm{H}$ lifetime puzzle theoretically using realistic three-body ${}_{\Lambda}^3\mathrm{H}$ and ${}^3\mathrm{He}$ wave functions computed within the ab initio no-core shell model approach with interactions derived from chiral effective field theory. In particular, $\tau({}_{\Lambda}^3\mathrm{H})$ was found to be strongly correlated with the $\Lambda$ separation energy $B_\Lambda$ in ${}_{\Lambda}^3\mathrm{H}$, the value of which is rather poorly known experimentally and, in addition, is known to suffer from sizable theoretical uncertainties inherent in the employed nuclear and hypernuclear interaction models. In the present work we find that these uncertainties propagate into $\tau({}_{\Lambda}^3\mathrm{H})$, and thus limit considerably the theoretical precision of its computed value. Although none of the conflicting RHI measured $\tau({}_{\Lambda}^3\mathrm{H})$ values can be excluded, but rather can be attributed to a poor knowledge of $B_\Lambda$, we note the good agreement between the lifetime value $\tau({}_{\Lambda}^3\mathrm{H})=238(27)$ ps computed at the lowest value $B_\Lambda=66$ keV reached by us and the very recent ALICE measured lifetime value $\tau^{\mathrm{ALICE}}({}_{\Lambda}^3\mathrm{H})=253(11)(6)$ ps associated with the ALICE measured $B_\Lambda$ value $B^{\mathrm{ALICE}}_\Lambda=102(63)(67)$ keV [Phys. Rev. Lett. 131, 102302 (2023)]., Comment: 15 pages, 6 figures, 3 tables; minor changes, matches published version

Quantum computers promise exponential speed ups over classical computers for various tasks. This emerging technology is expected to have its first huge impact in High Performance Computing (HPC), as it can solve problems beyond the reach of HPC. To that end, HPC will require quantum accelerators, which will enable applications to run on both classical and quantum devices, via hybrid quantum-classical nodes. Hybrid quantum-HPC applications should be scalable, executable on Quantum Error Corrected (QEC) devices, and could use quantum-classical primitives. However, the lack of scalability, poor performances, and inability to insert classical schemes within quantum applications has prevented current quantum frameworks from being adopted by the HPC community. This paper specifies the requirements of a hybrid quantum-classical framework compatible with HPC environments, and introduces a novel hardware-agnostic framework called Q-Pragma. This framework extends the classical programming language C++ heavily used in HPC via the addition of pragma directives to manage quantum computations.

Astrophysical Very-High-Energy (VHE, >10PeV) neutrinos deliver crucial information about the sources of Ultra-High-Energy Cosmic Rays (UHECRs), the composition of UHECRs, and neutrino/particle physics at highest energies. UHE-tau neutrinos skimming the Earth's surface produce tau leptons, which can emerge from the ground, decay, and start an upward-going extensive air shower (EAS) in the Earth's atmosphere. The tau neutrino can be reconstructed by imaging the EAS. We developed an atmospheric Cherenkov Telescope flying on the Extreme Universe Space Observatory Super Pressure Balloon 2 (EUSO-SPB2) mission to test the air-shower imaging concept at highest altitudes. The EUSO-SPB2 ultra-long-duration balloon mission is a precursor of the Probe of Extreme Multi-Messenger Astrophysics (POEMMA), a candidate for an astrophysics probe-class mission. The telescope implements Schmidt optics with a 0.785 m^2 light collection area and a 512-pixel SiPM camera covering a 12.8{\deg} by 6.4{\deg} (Horizontal by Vertical) field-of-view with 0.4{\deg} resolution. The camera signals are sampled with 100MSa/s and digitized with 12-bit resolution. The objectives of the EUSO-SPB2 Cherenkov telescope include a search for UHE neutrinos below Earth's limb, UHECRs above the limb, the study of the night sky background, and studying the telescope's performance. In this presentation, I will present an overview of the Cherenkov telescope and discuss the in-flight performance of the telescope.

The proliferation of technologies, such as extended reality (XR), has increased the demand for high-quality three-dimensional (3D) graphical representations. Industrial 3D applications encompass computer-aided design (CAD), finite element analysis (FEA), scanning, and robotics. However, current methods employed for industrial 3D representations suffer from high implementation costs and reliance on manual human input for accurate 3D modeling. To address these challenges, neural radiance fields (NeRFs) have emerged as a promising approach for learning 3D scene representations based on provided training 2D images. Despite a growing interest in NeRFs, their potential applications in various industrial subdomains are still unexplored. In this paper, we deliver a comprehensive examination of NeRF industrial applications while also providing direction for future research endeavors. We also present a series of proof-of-concept experiments that demonstrate the potential of NeRFs in the industrial domain. These experiments include NeRF-based video compression techniques and using NeRFs for 3D motion estimation in the context of collision avoidance. In the video compression experiment, our results show compression savings up to 48\% and 74\% for resolutions of 1920x1080 and 300x168, respectively. The motion estimation experiment used a 3D animation of a robotic arm to train Dynamic-NeRF (D-NeRF) and achieved an average peak signal-to-noise ratio (PSNR) of disparity map with the value of 23 dB and an structural similarity index measure (SSIM) 0.97.

It is often stated that the Carlitz module is to the ring of univariate polynomials over a finite field what the multiplicative group is to the ring of integers. This analogy extends to the "rank 2" case, where Drinfeld modules play a role similar to that of elliptic curves. This work grew out with the will of finding a common definition for these objects, depending only on the ring of coefficients, and thus elevating this analogy to a common theory. To that end, we introduce a class of algebraic $A$-modules for a finitely generated Dedekind ring $A$, called "modules \'el\'ementaires", which naturally generalize Drinfeld modules, forms of the multiplicative group, and elliptic curves over a field (when $A$ has the corresponding form). The objective of this text is the classification of these "modules \'el\'ementaires"., Comment: 21 pages, in French

Intensity modulated radiotherapy (IMRT) is one of the most common modalities for treating cancer patients. One of the biggest challenges is precise treatment delivery that accounts for varying motion patterns originating from free-breathing. Currently, image-guided solutions for IMRT is limited to 2D guidance due to the complexity of 3D tracking solutions. We propose a novel end-to-end attention graph neural network model that generates in real-time a triangular shape of the liver based on a reference segmentation obtained at the preoperative phase and a 2D MRI coronal slice taken during the treatment. Graph neural networks work directly with graph data and can capture hidden patterns in non-Euclidean domains. Furthermore, contrary to existing methods, it produces the shape entirely in a mesh structure and correctly infers mesh shape and position based on a surrogate image. We define two on-the-fly approaches to make the correspondence of liver mesh vertices with 2D images obtained during treatment. Furthermore, we introduce a novel task-specific identity loss to constrain the deformation of the liver in the graph neural network to limit phenomenons such as flying vertices or mesh holes. The proposed method achieves results with an average error of 3.06 +- 0.7 mm and Chamfer distance with L2 norm of 63.14 +- 27.28., Comment: Copyright 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works

The number of international benchmarking competitions is steadily increasing in various fields of machine learning (ML) research and practice. So far, however, little is known about the common practice as well as bottlenecks faced by the community in tackling the research questions posed. To shed light on the status quo of algorithm development in the specific field of biomedical imaging analysis, we designed an international survey that was issued to all participants of challenges conducted in conjunction with the IEEE ISBI 2021 and MICCAI 2021 conferences (80 competitions in total). The survey covered participants' expertise and working environments, their chosen strategies, as well as algorithm characteristics. A median of 72% challenge participants took part in the survey. According to our results, knowledge exchange was the primary incentive (70%) for participation, while the reception of prize money played only a minor role (16%). While a median of 80 working hours was spent on method development, a large portion of participants stated that they did not have enough time for method development (32%). 25% perceived the infrastructure to be a bottleneck. Overall, 94% of all solutions were deep learning-based. Of these, 84% were based on standard architectures. 43% of the respondents reported that the data samples (e.g., images) were too large to be processed at once. This was most commonly addressed by patch-based training (69%), downsampling (37%), and solving 3D analysis tasks as a series of 2D tasks. K-fold cross-validation on the training set was performed by only 37% of the participants and only 50% of the participants performed ensembling based on multiple identical models (61%) or heterogeneous models (39%). 48% of the respondents applied postprocessing steps.

The integral $t$-motivic cohomology and the class module of a (rigid analytically trivial) Anderson $t$-motive were introduced by the first author in [Gaz22b]. This paper is devoted to their determination in the particular case of tensor powers of the Carlitz $t$-motive, namely, the function field counterpart $\underline{A}(n)$ of Tate twists $\mathbb{Z}(n)$. We find out that these modules are in relation with fundamental objects of function field arithmetic: integral $t$-motivic cohomology governs linear relations among Carlitz polylogarithms, its torsion is expressed in terms of the denominator of Bernoulli-Carlitz numbers and the Fitting ideal of class modules is a special zeta value. We also express the regulator of $\underline{A}(n)$ for positive $n$ in terms of generalized Carlitz polylogarithms; after establishing their algebraic relations using difference Galois theory together with the Anderson-Brownawell-Papanikolas criterion, we prove that the regulator is an isomorphism if, and only if, $n$ is prime to the characteristic., Comment: This represents a significant update compared to the previous version. Notably, our late conjecture regarding the Fitting ideal of the torsion in the class module, expressed as Carlitz zeta values, has been established (Thm. C). Additionally, we clarified the connection between integral t-motivic cohomology and Carlitz polylogs (Thm. A) and determined the regulator ranks

Conflicting values of the hypertriton lifetime $\tau({}_\Lambda^3\mathrm{H})$ were derived in relativistic heavy ion (RHI) collision experiments over the last decade. A very recent ALICE Collaboration measurement is the only experiment where the reported $\tau({}_\Lambda^3\mathrm{H})$ comes sufficiently close to the free-$\Lambda$ lifetime $\tau_\Lambda$,as expected naively for a very weakly bound $\Lambda$ in ${}_\Lambda^3\mathrm{H}$. We revisited theoretically this ${}_\Lambda^3\mathrm{H}$ lifetime puzzle, using ${}_\Lambda^3\mathrm{H}$ and ${}^3\mathrm{He}$ wave functions computed within the abinitio no-core shell model employing interactions derived from chiral effective field theory to calculate the two-body decay rate $\Gamma({}_\Lambda^3\mathrm{H}\to{}^3\mathrm{He}+\pi^-)$. We found significant but opposing contributions arising from $\Sigma NN$ admixtures in ${}_\Lambda^3\mathrm{H}$ and from $\pi^- -{}^3\mathrm{He}$ final-state interaction. To derive $\tau({}_\Lambda^3\mathrm{H})$, we evaluated the inclusive $\pi^-$ decay rate $\Gamma_{\pi^-}({}_\Lambda^3\mathrm{H})$ by using the measured branching ratio $\Gamma({}_\Lambda^3\mathrm{H}\to{}^3\mathrm{He}+\pi^-)/\Gamma_{\pi^-}({}_\Lambda^3\mathrm{H})$ and added the $\pi^0$ contributions through the $\Delta I = \frac{1}{2}$ rule. The resulting $\tau({}_\Lambda^3\mathrm{H})$ varies strongly with the rather poorly known $\Lambda$ separation energy $E_{\mathrm{sep}}({}_\Lambda^3\mathrm{H})$ and it is thus possible to associate each one of the distinct RHI $\tau({}_\Lambda^3\mathrm{H})$ measurements with its own underlying value of $E_{\mathrm{sep}}({}_\Lambda^3\mathrm{H})$., Comment: Invited talk given at the 14th International Conference on Hypernuclear and Strange Particle Physics (HYP2022), Prague, Czech Republic, June 27 - July 1, 2022. A slightly updated version of the published proceedings contribution

Let $E$ be an Anderson $A$-module over $\mathbb{C}_{\infty}$. The period lattice of $E$ is related to its module of special functions by means of a non-canonical isomorphism introduced by the authors in [GM21]. In this paper, we explain how a modification of the inverse map is canonical by interpreting it as a residue morphism along the characteristic graph. This phenomenon has already been observed in various situations. The main innovation of this text is that of costability (costable admissible opens, costable site, etc.) which provides a convenient framework to develop the notion of sheaves of $E(\mathbb{C}_{\infty})$-valued meromorphic functions on the rigid analytic plane., Comment: 23 pages. Final version. To appear in Journal of Number Theory

Recent trends in robotic services propose offloading robot functionalities to the Edge to meet the strict latency requirements of networked robotics. However, the Edge is typically an expensive resource and sometimes the Cloud is also an option, thus, decreasing the cost. Following this idea, we propose Don't Let Me Down! (DLMD), an algorithm that promotes offloading robot functions to the Cloud when possible to minimize the consumption of Edge resources. Additionally, DLMD takes the appropriate migration, traffic steering, and radio handover decisions to meet robotic service requirements as strict latency constraints. In the paper, we formulate the optimization problem that DLMD aims to solve, compare DLMD performance against state of art, and perform stress tests to assess DLMD performance in small & large networks. Results show that DLMD (i) always finds solutions in less than 30ms; (ii) is optimal in a local warehousing use case, and (iii) consumes only 5% of the Edge resources upon network stress., Comment: 5 Pages, 6 figures, submitted to 2023 IEEE 9th International Conference on Network Softwarization (NetSoft)

High entropy oxides are emerging as an exciting new avenue to design highly tailored functional behaviors that have no traditional counterparts. Study and application of these materials are bringing together scientists and engineers from physics, chemistry, and materials science. The diversity of each of these disciplines comes with perspectives and jargon that may be confusing to those outside of the individual fields, which can result in miscommunication of important aspects of research. In this perspective, we provide examples of research and characterization taken from these different fields to provide a framework for classifying the differences between compositionally complex oxides, high entropy oxides, and entropy stabilized oxides, which is intended to bring a common language to this emerging area. We highlight the critical importance of understanding a materials crystallinity, composition, and mixing length scales in determining its true definition.

The kidney cancer is one of the most common cancer types. The treatment frequently include surgical intervention. However, surgery is in this case particularly challenging due to regional anatomical relations. Organ delineation can significantly improve surgical planning and execution. In this contribution, we propose ensemble of two fully convolutional networks for segmentation of kidney, tumor, veins and arteries. While SegResNet architecture achieved better performance on tumor, the nnU-Net provided more precise segmentation for kidneys, arteries and veins. So in our proposed approach we combine these two networks, and further boost the performance by mixup augmentation.

The energy levels of light hypernuclei are experimentally accessible observables that contain valuable information about the interaction between hyperons and nucleons. In this work we study strangeness $S = -1$ systems $^{3,4}_\Lambda$H and $^{4,5}_\Lambda$He using the ab initio no-core shell model (NCSM) with realistic interactions obtained from chiral effective field theory ($\chi$EFT). In particular, we quantify the finite precision of theoretical predictions that can be attributed to nuclear physics uncertainties. We study both the convergence of the solution of the many-body problem (method uncertainty) and the regulator- and calibration data-dependence of the nuclear $\chi$EFT Hamiltonian (model uncertainty). For the former, we implement infrared correction formulas and extrapolate finite-space NCSM results to infinite model space. We then use Bayesian parameter estimation to quantify the resulting method uncertainties. For the latter, we employ a family of 42 realistic Hamiltonians and measure the standard deviation of predictions while keeping the leading-order hyperon-nucleon interaction fixed. Following this procedure we find that model uncertainties of ground-state $\Lambda$ separation energies amount to $\sim 20(100)$ keV in $^3_\Lambda$H($^4_\Lambda$H,He) and $\sim 400$ keV in $^5_\Lambda$He. Method uncertainties are comparable in magnitude for the $^4_\Lambda$H,He $1^+$ excited states and $^5_\Lambda$He, which are computed in limited model spaces, but otherwise much smaller. This knowledge of expected theoretical precision is crucial for the use of binding energies of light hypernuclei to infer the elusive hyperon-nucleon interaction., Comment: 16 pages with 8 figures

As a natural sequel for the study of $A$-motivic cohomology, initiated in [Gaz], we develop a notion of regulator for rigid analytically trivial Anderson $A$-motives. In accordance with the conjectural number field picture, we define it as the morphism at the level of extension modules induced by the exactness of the Hodge-Pink realization functor. The purpose of this text is twofold: we first prove a finiteness result for $A$-motivic cohomology and, under a weight assumption, we then show that the source and the target of regulators have the same dimension. It came as a surprise to the author that the image of this regulator might not have full rank, preventing the analogue of a renowned conjecture of Beilinson to hold in our setting., Comment: v1->v2: Corrections: removed the mixedness assumption; changed some notations to improve readability; added details; corrected typos

The challenge of dynamic traffic demand in mobile networks is tackled by moving cells based on unmanned aerial vehicles. Considering the tremendous potential of unmanned aerial vehicles in the future, we propose a new heuristic algorithm for coverage optimization. The proposed algorithm is implemented based on a conditional generative adversarial neural network, with a unique multilayer sum-pooling loss function. To assess the performance of the proposed approach, we compare it with the optimal core-set algorithm and quasi-optimal spiral algorithm. Simulation results show that the proposed approach converges to the quasi-optimal solution with a negligible difference from the global optimum while maintaining a quadratic complexity regardless of the number of users., Comment: International Journal of Distributed Sensor Networks. 2022

Compilation of papers presented by the JEM-EUSO Collaboration at the 37th International Cosmic Ray Conference (ICRC), held on July 12-23, 2021 (online) in Berlin, Germany., Comment: html page with links to the JEM-EUSO Collaboration papers presented at ICRC-2021, Berlin, Germany

The deepest arithmetic invariants attached to an algebraic variety defined over a number field $F$ are conjecturally captured by the integral part of its motivic cohomology. There are essentially two ways of defining it when $X$ is a smooth projective variety: one is via the $K$-theory of a regular model, the other is through its $\ell$-adic realization. Both approaches are conjectured to coincide. This paper initiates the study of motivic cohomology for global fields of positive characteristic, hereafter named $A$-motivic cohomology, where classical mixed motives are replaced by mixed Anderson $A$-motives. Our main objective is to set the definitions of the model version and the $\ell$-adic version of the integral part of $A$-motivic cohomology, using Gardeyn's notion of maximal models of $A$-motives as the analogue of regular models of varieties. Our main result states that the model version is contained in the $\ell$-adic version. As opposed to what is expected in the number field setting, we show that the two approaches do not match in general. We conclude this work by introducing the submodule of regulated extensions of mixed Anderson $A$-motives, for which we expect the two approaches to match, and solve some particular cases of this expectation., Comment: 61 pages. Final version. To appear in Compositio Mathematica

Chest radiography is a relatively cheap, widely available medical procedure that conveys key information for making diagnostic decisions. Chest X-rays are almost always used in the diagnosis of respiratory diseases such as pneumonia or the recent COVID-19. In this paper, we propose a self-supervised deep neural network that is pretrained on an unlabeled chest X-ray dataset. The learned representations are transferred to downstream task - the classification of respiratory diseases. The results obtained on four public datasets show that our approach yields competitive results without requiring large amounts of labeled training data., Comment: The work was published by IEEE Access. DOI: 10.1109/ACCESS.2021.3125324