This work analyzes the use of large language models (LLMs) for detecting domain generation algorithms (DGAs). We perform a detailed evaluation of two important techniques: In-Context Learning (ICL) and Supervised Fine-Tuning (SFT), showing how they can improve detection. SFT increases performance by using domain-specific data, whereas ICL helps the detection model to quickly adapt to new threats without requiring much retraining. We use Meta's Llama3 8B model, on a custom dataset with 68 malware families and normal domains, covering several hard-to-detect schemes, including recent word-based DGAs. Results proved that LLM-based methods can achieve competitive results in DGA detection. In particular, the SFT-based LLM DGA detector outperforms state-of-the-art models using attention layers, achieving 94% accuracy with a 4% false positive rate (FPR) and excelling at detecting word-based DGA domains.

During the last years, algorithms known as Convolutional Neural Networks (CNNs) had become increasingly popular, expanding its application range to several areas. In particular, the image processing field has experienced a remarkable advance thanks to this algorithms. In IoT, a wide research field aims to develop hardware capable of execute them at the lowest possible energy cost, but keeping acceptable image inference time. One can get around this apparently conflicting objectives by applying design and training techniques. The present work proposes a generic hardware architecture ready to be implemented on FPGA devices, supporting a wide range of configurations which allows the system to run different neural network architectures, dynamically exploiting the sparsity caused by pruning techniques in the mathematical operations present in this kind of algorithms. The inference speed of the design is evaluated over different resource constrained FPGA devices. Finally, the standard pruning algorithm is compared against a custom pruning technique specifically designed to exploit the scheduling properties of this hardware accelerator. We demonstrate that our hardware-aware pruning algorithm achieves a remarkable improvement of a 45 % in inference time compared to a network pruned using the standard algorithm.

In this paper, we consider a k-nearest neighbor kernel type estimator when the random variables belong in a Riemannian manifolds. We study asymptotic properties such as the consistency and the asymptotic distribution. A simulation study is also consider to evaluate the performance of the proposal. Finally, to illustrate the potential applications of the proposed estimator, we analyzed two real example where two different manifolds are considered.

Ground-based gamma-ray astronomy is now well established as a key observational approach to address critical topics at the frontiers of astroparticle physics and high-energy astrophysics. Whilst the field of TeV astronomy was once dominated by arrays of atmospheric Cherenkov Telescopes, ground-level particle detection has now been demonstrated to be an equally viable and strongly complementary approach. Ground-level particle detection provides continuous monitoring of the overhead sky, critical for the mapping of extended structures and capturing transient phenomena. As demonstrated by HAWC and LHAASO, the technique provides the best available sensitivity above a few tens of TeV, and for the first time access to the PeV energy range. Despite the success of this approach, there is so far no major ground-level particle-based observatory with access to the Southern sky. HESS, located in Namibia, is the only major gamma-ray instrument in the Southern Hemisphere, and has shown the extraordinary richness of the inner galaxy in the TeV band, but is limited in terms of field of view and energy reach. SWGO is an international effort to construct the first wide-field instrument in the south with deep sensitivity from 100s of GeV into the PeV domain. The project is now close to the end of its development phase and planning for construction of the array in Chile has begun. Here we describe the baseline design, expected sensitivity and resolution, and describe in detail the main scientific topics that will be addressed by this new facility and its initial phase SWGO-A. We show that SWGO will have a transformational impact on a wide range of topics from cosmic-ray acceleration and transport to the nature of dark matter. SWGO represents a key piece of infrastructure for multi-messenger astronomy in the next decade, with strong scientific synergies with the nearby CTA Observatory.

The development of successful artificial intelligence models for chest X-ray analysis relies on large, diverse datasets with high-quality annotations. While several databases of chest X-ray images have been released, most include disease diagnosis labels but lack detailed pixel-level anatomical segmentation labels. To address this gap, we introduce an extensive chest X-ray multi-center segmentation dataset with uniform and fine-grain anatomical annotations for images coming from five well-known publicly available databases: ChestX-ray8, Chexpert, MIMIC-CXR-JPG, Padchest, and VinDr-CXR, resulting in 657,566 segmentation masks. Our methodology utilizes the HybridGNet model to ensure consistent and high-quality segmentations across all datasets. Rigorous validation, including expert physician evaluation and automatic quality control, was conducted to validate the resulting masks. Additionally, we provide individualized quality indices per mask and an overall quality estimation per dataset. This dataset serves as a valuable resource for the broader scientific community, streamlining the development and assessment of innovative methodologies in chest X-ray analysis. The CheXmask dataset is publicly available at: this https URL

We report a measurement of the energy spectrum of cosmic rays for energies above $2.5 {\times} 10^{18}~$eV based on 215,030 events recorded with zenith angles below $60^\circ$. A key feature of the work is that the estimates of the energies are independent of assumptions about the unknown hadronic physics or of the primary mass composition. The measurement is the most precise made hitherto with the accumulated exposure being so large that the measurements of the flux are dominated by systematic uncertainties except at energies above $5 {\times} 10^{19}~$eV. The principal conclusions are: (1) The flattening of the spectrum near $5 {\times} 10^{18}~$eV, the so-called "ankle", is confirmed. (2) The steepening of the spectrum at around $5 {\times} 10^{19}~$eV is confirmed. (3) A new feature has been identified in the spectrum: in the region above the ankle the spectral index $\gamma$ of the particle flux ($\propto E^{-\gamma}$) changes from $2.51 \pm 0.03~{\rm (stat.)} \pm 0.05~{\rm (sys.)}$ to $3.05 \pm 0.05~{\rm (stat.)} \pm 0.10~{\rm (sys.)}$before changing sharply to$5.1 \pm 0.3~{\rm (stat.)} \pm 0.1~{\rm (sys.)}$above$5 {\times} 10^{19}~$eV. (4) No evidence for any dependence of the spectrum on declination has been found other than a mild excess from the Southern Hemisphere that is consistent with the anisotropy observed above $8 {\times} 10^{18}~$eV.

Variable selection problems generally present more than a single solution and, sometimes, it is worth to find as many solutions as possible. The use of Evolutionary Algorithms applied to this kind of problem proves to be one of the best methods to find optimal solutions. Moreover, there are variants designed to find all or almost all local optima, known as Niching Genetic Algorithms (NGA). There are several different NGA methods developed in order to achieve this task. The present work compares the behavior of eight different niching techniques, applied to a climatic database of four weather stations distributed in Tucuman, Argentina. The goal is to find different sets of input variables that have been used as the input variable by the estimation method. Final results were evaluated based on low estimation error and low dispersion error, as well as a high number of different results and low computational time. A second experiment was carried out to study the capability of the method to identify critical variables. The best results were obtained with Deterministic Crowding. In contrast, Steady State Worst Among Most Similar and Probabilistic Crowding showed good results but longer processing times and less ability to determine the critical factors.

A search for ultra-high energy photons with energies above 1 EeV is performed using nine years of data collected by the Pierre Auger Observatory in hybrid operation mode. An unprecedented separation power between photon and hadron primaries is achieved by combining measurements of the longitudinal air-shower development with the particle content at ground measured by the fluorescence and surface detectors, respectively. Only three photon candidates at energies 1 - 2 EeV are found, which is compatible with the expected hadron-induced background. Upper limits on the integral flux of ultra-high energy photons of 0.038, 0.010, 0.009, 0.008 and 0.007 km$^{-2}$ sr$^{-1}$ yr$^{-1}$ are derived at 95% C.L. for energy thresholds of 1, 2, 3, 5 and 10 EeV. These limits bound the fractions of photons in the all-particle integral flux below 0.14%, 0.17%, 0.42%, 0.86% and 2.9%. For the first time the photon fraction at EeV energies is constrained at the sub-percent level. The improved limits are below the flux of diffuse photons predicted by some astrophysical scenarios for cosmogenic photon production. The new results rule-out the early top-down models $-$ in which ultra-high energy cosmic rays are produced by, e.g., the decay of super-massive particles $-$ and challenge the most recent super-heavy dark matter models.

This research presents the development of a new simulation model to determine the optimal order lot sizes in Material Requirements Planning, based on purchase volume and the temporal deterioration of items. The scientific novelty lies in the exhaustive enumeration of all supply strategies, considering when and how much raw material and/or inputs to acquire while simultaneously managing multiple factors. The developed model enables obtaining and visualizing all feasible solutions, though a significant challenge arises when dealing with larger planning horizons. The methodology includes formulating a mathematical equation to calculate the total cost of all supply strategies, taking into account quantity discounts and the maximum allowable shelf life of inventory items. The results demonstrate that the model thoroughly explores the entire search space and identifies the optimal solution. Validation is conducted using the tabu search heuristic, a widely recognized optimization technique. While the heuristic converges toward the global minimum, it requires a significantly higher computational load. In contrast, the developed model identifies the global minimum with fewer calculations, showcasing its efficiency and accuracy.

The obstructive sleep apnea-hypopnea (OSAH) syndrome is a very common and frequently undiagnosed sleep disorder. It is characterized by repeated events of partial (hypopnea) or total (apnea) obstruction of the upper airway while sleeping. This study makes use of a previously developed method called DAS-KSVD for multiclass structured dictionary learning to automatically detect individual events of apnea and hypopnea using only blood oxygen saturation signals. The method uses a combined discriminant measure which is capable of efficiently quantifying the degree of discriminability of each one of the atoms in a dictionary. DAS-KSVD was applied to detect and classify apnea and hypopnea events from signals obtained from the Sleep Heart Health Study database. For moderate to severe OSAH screening, a receiver operating characteristic curve analysis of the results shows an area under the curve of 0.957 and diagnostic sensitivity and specificity of 87.56% and 88.32%, respectively. These results represent improvements as compared to most state-of-the-art procedures. Hence, the method could be used for screening OSAH syndrome more reliably and conveniently, using only a pulse oximeter.

This essay presents an exploration of elements from information theory and cibernetics on the struggle against corruption behavior in public sector and beyond; the existence of an exemplary or corrupt ethical equilibriums are explored by updating Klitgaard corruption formula along with the presence of information pressure, entropy and cibernetics servomechanisms in digital societies, including alternatives and sistemics approaches for further anti-corruption policies implementation.

Inverse imaging problems that are ill-posed can be encountered across multiple domains of science and technology, ranging from medical diagnosis to astronomical studies. To reconstruct images from incomplete and distorted data, it is necessary to create algorithms that can take into account both, the physical mechanisms responsible for generating these measurements and the intrinsic characteristics of the images being analyzed. In this work, the sparse representation of images is reviewed, which is a realistic, compact and effective generative model for natural images inspired by the visual system of mammals. It enables us to address ill-posed linear inverse problems by training the model on a vast collection of images. Moreover, we extend the application of sparse coding to solve the non-linear and ill-posed problem in microwave tomography imaging, which could lead to a significant improvement of the state-of-the-arts algorithms.

We present the Python-based Molecule Builder for ESPResSo (pyMBE), an open source software to design custom Coarse-Grained (CG) models, as well as pre-defined models of polyelectrolytes, peptides and globular proteins in the Extensible Simulation Package for Research on Soft Matter (ESPResSo). The Python interface of \espresso offers a flexible framework, capable of building custom CG models from scratch. As a downside, building CG models from scratch is error-prone, especially for newcomers in the field of CG modeling, or for molecules with complex architectures. The pyMBE module builds CG models in \espresso using a hierarchical bottom-up approach, providing a robust tool to automate the setup of CG models and helping new users prevent common mistakes. ESPResSo features the constant pH (cpH) and grand-reaction (G-RxMC) methods, which have been designed to study chemical reaction equilibria in macromolecular systems with many reactive species. However, setting up these methods for systems which contain several types of reactive groups is an error-prone task, especially for beginners. The pyMBE module enables the automatic setup of cpH and G-RxMC simulations in \espresso, lowering the barrier for newcomers and opening the door to investigate complex systems not studied with these methods yet. To demonstrate some of the applications of pyMBE, we showcase several case studies where we successfully reproduce previously published simulations of charge-regulating peptides and globular proteins in bulk solution and weak polyelectrolytes in dialysis. The pyMBE module is publicly available as a GitHub repository (this https URL) which includes its source code and various sample and test scripts, including the ones that we used to generate the data presented in this article.

The convergence between astronomy and data sonification represents a significant advancement in the approach and analysis of cosmic information. By surpassing the visual exclusivity in data analysis in astronomy, innovative projects have developed software that goes beyond visual representation, transforming data into auditory and tactile displays. However, it has been evidenced that this novel technique requires specialized training, particularly for audio format data. This work describes the initial development of a platform aimed at providing training for data analysis in astronomy through sonification. The integration of these tools in astronomical education and research opens new horizons, facilitating a more inclusive and multisensory participation in the exploration of space science.

The distinguishability between two quantum states can be defined in terms of their trace distance. The operational meaning of this definition involves a maximization over measurement projectors. Here we introduce an alternative definition of distinguishability which, instead of projectors, is based on maximization over normalized states (density matrices). It is shown that this procedure leads to a distance (between two states) that, in contrast to the usual approach based on a 1-norm, is based on an infinite-norm. Properties such as convexity, monotonicity, and invariance under unitary transformations are fulfilled. Equivalent operational implementations based on maximization over classical probabilities and hypothesis testing scenarios are also established. When considering the action of completely positive transformations contractivity is only granted for unital maps. This feature allows us to introduce a measure of the quantumness of non-unital maps that can be written in terms of the proposed distinguishability measure and corresponds to the maximal possible deviation from contractivity. Particular examples sustain the main results and conclusions.

We characterize a class of superclassical non-Markovian open quantum system dynamics that are defined by their lack of measurement invasiveness when the corresponding observable commutates with the pre-measurement state. This diagonal non-invasiveness guarantees that joint probabilities for measurement outcomes fulfill classical Kolmogorov consistency conditions. These features are fulfilled regardless of the previous (measurement) system history and are valid at arbitrary later times after an arbitrary system initialization. It is shown that a subclass of depolarizing dynamics, which are based on a (time-irreversible) non-unitary system-environment coupling, satisfy the required properties. The relationship with other operational [Milz et al., Phys. Rev. X 10, 041049 (2020)] and non-operational [Banacki et al., Phys. Rev. A 107, 032202 (2023)] notions of classicality in non-Markovian open quantum systems is studied in detail and exemplified through different examples.

The growing need for companies to reduce costs and maximize profits has led to an increased focus on logistics activities. Among these, inventory management plays a crucial role in minimizing organizational expenses by optimizing the storage and transportation of materials. In this context, this study introduces an optimization model for the lot-sizing problem based on a physical system approach. By establishing that the material supply problem is isomorphic to a one-dimensional mechanical system of point particles connected by elastic elements, we leverage this analogy to derive cost optimization conditions naturally and obtain an exact solution. This approach determines lot sizes that minimize the combined ordering and inventory holding costs in a significantly shorter time, eliminating the need for heuristic methods. The optimal lot sizes are defined in terms of the parameter $ \gamma = 2C_O / C_H $, which represents the relationship between the ordering cost per order ($ C_O $) and the holding cost per period for the material required in one period ($ C_H $). This parameter fully dictates the system's behavior: when$ \gamma \leq 1 $, the optimal strategy is to place one order per period, whereas for$ \gamma > 1 $, the number of orders$ N $ is reduced relative to the planning horizon $M$, meaning N < M . By formulating the total cost function in terms of the intensive variable $N/M$, we consolidate the entire optimization problem into a single function of $\gamma$. This eliminates the need for complex algorithms, enabling faster and more precise purchasing decisions. The proposed model was validated through a real-world case study and benchmarked against classical algorithms, demonstrating superior cost optimization and reduced execution time. These findings underscore the potential of this approach for improving material lot-sizing strategies.

The disruption of protein structures by denaturants like urea is well studied, though its molecular mechanisms remain unclear. Using Molecular Dynamics (MD) simulations, we investigated how urea affects the structural stability of Bovine Serum Albumin (BSA) at concentrations from 0 M to 5 M. Our results reveal that urea induces a dehydration/rehydration cycle, characterized by displacement and partial replacement of water molecules in BSAs hydration shell. At low concentrations, urea reduces protein/water hydrogen bonds while enhancing protein-urea interactions. At higher concentrations, urea aggregation limits these interactions, promoting rehydration and changes in tertiary structure, while secondary structure remains largely intact. These findings provide insights into the mechanisms of protein denaturation and stability by urea.

The photovoltaic industry faces the challenge of optimizing the performance and management of its systems in an increasingly digitalized environment. In this context, digital twins offer an innovative solution: virtual models that replicate in real time the behavior of solar installations. This technology makes it possible to anticipate failures, improve operational efficiency and facilitate data-driven decision-making. This report analyzes its application in the photovoltaic sector, highlighting its benefits and transformative potential.

With the increase of machine learning usage by industries and scientific communities in a variety of tasks such as text mining, image recognition and self-driving cars, automatic setting of hyper-parameter in learning algorithms is a key factor for achieving satisfactory performance regardless of user expertise in the inner workings of the techniques and methodologies. In particular, for a reinforcement learning algorithm, the efficiency of an agent learning a control policy in an uncertain environment is heavily dependent on the hyper-parameters used to balance exploration with exploitation. In this work, an autonomous learning framework that integrates Bayesian optimization with Gaussian process regression to optimize the hyper-parameters of a reinforcement learning algorithm, is proposed. Also, a bandits-based approach to achieve a balance between computational costs and decreasing uncertainty about the Q-values, is presented. A gridworld example is used to highlight how hyper-parameter configurations of a learning algorithm (SARSA) are iteratively improved based on two performance functions.