Search

This paper describes our participation in the MentalRiskES task at IberLEF 2023. The task involved predicting the likelihood of an individual experiencing depression based on their social media activity. The dataset consisted of conversations from 175 Telegram users, each labeled according to their evidence of suffering from the disorder. We used a combination of traditional machine learning and deep learning techniques to solve four predictive subtasks: binary classification, simple regression, multiclass classification, and multi-output regression. We approached this by training a model to solve the multi-output regression case and then transforming the predictions to work for the other three subtasks. We compare the performance of two modeling approaches: fine-tuning a BERT-based model directly for the task or using its embeddings as inputs to a linear regressor, with the latter yielding better results. The code to reproduce our results can be found at: https://github.com/simonsanvil/EarlyDepression-MentalRiskES, Comment: Submitted to the Proceedings of IberLEF 2023, September 2023, Ja\'en, Spain

In this work, we propose and demonstrate experimentally a new family of vector beams, the Helico-Conical Vector Beams (HCVB), whose spatial degree of freedom is encoded in the Helico-Conical Optical Beams. We use Stokes polarimetry to study their properties and find that upon propagation their transverse polarisation distribution evolves from non-homogeneous to quasi-homogeneous, such that even though their global degree of nonseparability remains constant, locally it decreases to a minimum value as z tends to infinit. We corroborated this quantitatively using the Hellinger distance, a novel metric for vectorness that applies to spatially-disjoint vector modes. To the best of our knowledge, HCVBs are the second family of vector beams in which this behaviour has been observed, paving the way for applications in optical tweezing or information encryption., Comment: 4 pages, 4 figures

New asymptotic approximations of the non-central $t$ distribution are given, a generalization of the Student's $t$ distribution. Using new integral representations, we give new asymptotic expansions for large values of the noncentrality parameter but also for large values of the degrees of freedom parameter. In some case we accept more than one large parameter. These results are in terms of elementary functions, but also in terms of the complementary error function and the incomplete gamma function. A number of numerical tests demonstrate the performance of the asymptotic approximations., Comment: 31 pages, 2 figures

We present a polarization variability analysis of a sample of 26 $\gamma$-ray blazars monitored by the Steward Observatory between 2008 and 2018 in the optical band. We investigate the properties and long-term variability of their optical polarization, searching for differences between blazar types. We observe that BL Lac objects are typically less polarized and less variable than flat spectrum radio quasars (FSRQs). Moreover, BL Lacs display a distribution of their polarization angle typically oriented in a preferential direction, contrary to the rather random distribution of FSRQs. For the latter blazar type, as well as those sources showing a bright stellar emission, we take into account the depolarizing effect introduced by the broad line region and the host galaxy on the measured polarization degree. In this sample we also observe that BL Lacs present an uncorrelated evolution of the flux and the polarization. Contrary, FSRQs show a correlation before the depolarization correction, that is lost however after considering this effect. In addition, we study the behaviour of the polarization angle, searching for angle rotations in its long-term evolution. We derive that the FSRQs studied here show rotations more frequently than BL Lac objects by a factor $\sim$1.5. During these periods we also observe a systematic decrease of the polarization fraction, as well as a marginal flux increase, not significant however to connect rotations with optical flares. We interpret these results within the extended shock-in-jet scenario, able to explain the overall features observed here for the polarization of the blazar sample., Comment: Accepted for publication in MNRAS

Low-dimensional quantum systems can host anyons, particles with exchange statistics that are neither bosonic nor fermionic. Despite indications of a wealth of exotic phenomena, the physics of anyons in one dimension (1D) remains largely unexplored. Here, we realize Abelian anyons in 1D with arbitrary exchange statistics using ultracold atoms in an optical lattice, where we engineer the statistical phase via a density-dependent Peierls phase. We explore the dynamical behavior of two anyons undergoing quantum walks, and observe the anyonic Hanbury Brown-Twiss effect, as well as the formation of bound states without on-site interactions. Once interactions are introduced, we observe spatially asymmetric transport in contrast to the symmetric dynamics of bosons and fermions. Our work forms the foundation for exploring the many-body behavior of 1D anyons., Comment: 8+10 pages, 5 figures

In an era of widespread web scraping, unlearnable dataset methods have the potential to protect data privacy by preventing deep neural networks from generalizing. But in addition to a number of practical limitations that make their use unlikely, we make a number of findings that call into question their ability to safeguard data. First, it is widely believed that neural networks trained on unlearnable datasets only learn shortcuts, simpler rules that are not useful for generalization. In contrast, we find that networks actually can learn useful features that can be reweighed for high test performance, suggesting that image protection is not assured. Unlearnable datasets are also believed to induce learning shortcuts through linear separability of added perturbations. We provide a counterexample, demonstrating that linear separability of perturbations is not a necessary condition. To emphasize why linearly separable perturbations should not be relied upon, we propose an orthogonal projection attack which allows learning from unlearnable datasets published in ICML 2021 and ICLR 2023. Our proposed attack is significantly less complex than recently proposed techniques., Comment: Accepted to NeurIPS 2023. Code available at https://github.com/psandovalsegura/learn-from-unlearnable

Four different relations between the Legendre nodes and weights are presented which, unlike the circle and trapezoid theorems for Gauss-Legendre quadrature, hold uniformly in the whole interval of orthogonality $(-1,1)$. These properties are supported by strong asymptotic evidence. The study of these results was originally motivated by the role some of them play in certain finite difference schemes used in the discretization of the angular Fokker-Planck diffusion operator., Comment: Section 3.1 has been added to this new version

Accreting low mass X-ray binaries (LMXBs) are capable of launching powerful outflows such as accretion disc winds. In disc winds, vast amounts of material can be carried away, potentially greatly impacting the binary and its environment. Previous studies have uncovered signatures of disc winds in the X-ray, optical, near-infrared, and recently even the UV band, predominantly in LMXBs with large discs ($P_{orb}{\geq}20$ hrs). Here, we present the discovery of transient UV outflow features in UW CrB, a high-inclination ($i{\geq}77$\deg) neutron star LMXB with an orbital period of only $P_{orb}{\approx}111$ min. We present P-Cygni profiles detected for Si iv 1400\r{A} and tentatively for N v 1240\r{A} in one 15 min exposure, which is the only exposure covering orbital phase $\phi{\approx}0.7{-}0.8$, with a velocity of ${\approx}1500$ km/s. We show that due to the presence of black body emission from the neutron star surface and/or boundary layer, a thermal disc wind can be driven despite the short $P_{orb}$, but explore alternative scenarios as well. The discovery that thermal disc winds may occur in NS-LMXBs with $P_{orb}$ as small as ${\approx}111$ min, and can potentially be transient on time scales as short as ${\approx}15$ min, warrants further observational and theoretical work., Comment: 5 pages, 5 figures. Submitted to MNRAS

GaGeTe is a layered topological semimetal that has been recently found to exist in at least two different polytypes, $\alpha$-GaGeTe ($R\bar{3}m$) and $\beta$-GaGeTe ($P6_3 mc$). Here we report a joint experimental and theoretical study of the structural, vibrational, and electronic properties of these two polytypes at high pressure. Both polytypes show anisotropic compressibility and two phase transitions, above 7 and 15 GPa, respectively, as confirmed by XRD and Raman spectroscopy measurements. Although the nature of the high-pressure phases is not confirmed, comparison with other chalcogenides and total-energy calculations allow us to propose possible high-pressure phases for both polytypes with an increase in coordination for Ga and Ge atoms from 4 to 6. In particular, the simplification of the X-ray patterns for both polytypes above 15 GPa suggests a transition to a structure of relatively higher symmetry than the original one. This result is consistent with the rocksalt-like high-pressure phases observed in parent III-VI semiconductors, such as GaTe, GaSe, and InSe. Pressure-induced amorphization is observed upon pressure release. The electronic band structures of $\alpha$-GaGeTe and $\beta$-GaGeTe and their pressure dependence also show similarities to III-VI semiconductors, thus suggesting that the germanene-like sublayer induces a semimetallic character in both GaGeTe polytypes. Above 3 GPa, both polytypes lose their topological features, due to the opening of the direct band gap, while the reduction of the interlayer space increases the thermal conductivity at high pressure., Comment: Main manuscript: 47 pages, 12 figures, 4 tables; ESI: 20 pages, 12 figures, 9 tables

Numerous observations suggest that there exist undiscovered beyond-the-Standard-Model particles and fields. Because of their unknown nature, these exotic particles and fields could interact with Standard Model particles in many different ways and assume a variety of possible configurations. Here we present an overview of the Global Network of Optical Magnetometers for Exotic physics searches (GNOME), our ongoing experimental program designed to test a wide range of exotic physics scenarios. The GNOME experiment utilizes a worldwide network of shielded atomic magnetometers (and, more recently, comagnetometers) to search for spatially and temporally correlated signals due to torques on atomic spins from exotic fields of astrophysical origin. We survey the temporal characteristics of a variety of possible signals currently under investigation such as those from topological defect dark matter (axion-like particle domain walls), axion-like particle stars, solitons of complex-valued scalar fields (Q-balls), stochastic fluctuations of bosonic dark matter fields, a solar axion-like particle halo, and bursts of ultralight bosonic fields produced by cataclysmic astrophysical events such as binary black hole mergers., Comment: 22 pages, 12 figures, submitted to Annalen der Physik

State-of-the-art congestion control algorithms for data centers alone do not cope well with transient congestion and high traffic bursts. To help with these, we revisit the concept of direct \emph{backward} feedback from switches and propose Back-to-Sender (BTS) signaling to many concurrent incast senders. Combining it with our novel approach to in-network caching, we achieve near-source sub-RTT congestion signaling. Source Flow Control (SFC) combines these two simple signaling mechanisms to instantly pause traffic sources, hence avoiding the head-of-line blocking problem of conventional hop-by-hop flow control. Our prototype system and scale simulations demonstrate that near-source signaling can significantly reduce the message completion time of various workloads in the presence of incast, complementing existing congestion control algorithms. Our results show that SFC can reduce the $99^{th}$-percentile flow completion times by $1.2-6\times$ and the peak switch buffer usage by $2-3\times$ compared to the recent incast solutions.

Generative models are a powerful tool in AI for material discovery. We are designing a software framework that supports a human-AI co-creation process to accelerate finding replacements for the ``forever chemicals''-- chemicals that enable our modern lives, but are harmful to the environment and the human health. Our approach combines AI capabilities with the domain-specific tacit knowledge of subject matter experts to accelerate the material discovery. Our co-creation process starts with the interaction between the subject matter experts and a generative model that can generate new molecule designs. In this position paper, we discuss our hypothesis that these subject matter experts can benefit from a more iterative interaction with the generative model, asking for smaller samples and ``guiding'' the exploration of the discovery space with their knowledge., Comment: 5 pages, Generative AI and HCI (GenAICHI) Workshop at CHI 23 (ACM CHI Conference on Human Factors in Computing Systems)

Recently developed text-to-image diffusion models make it easy to edit or create high-quality images. Their ease of use has raised concerns about the potential for malicious editing or deepfake creation. Imperceptible perturbations have been proposed as a means of protecting images from malicious editing by preventing diffusion models from generating realistic images. However, we find that the aforementioned perturbations are not robust to JPEG compression, which poses a major weakness because of the common usage and availability of JPEG. We discuss the importance of robustness for additive imperceptible perturbations and encourage alternative approaches to protect images against editing., Comment: 8 pages, 8 figures

We deal with existence and uniqueness of nonnegative solutions to \begin{equation*} \left\{ \begin{array}{l} -\Delta u = f(x) \text{ in }\Omega, \frac{\partial u}{\partial \nu} + \lambda(x) u = \frac{g(x)}{u^\eta} \text{ on } \partial\Omega, \end{array} \right. \end{equation*} where $\eta\ge 0$ and $f,\lambda$ and $g$ are nonnegative integrable functions. The set $\Omega\subset\mathbb{R}^N (N> 2)$ is open and bounded with smooth boundary and $\nu$ denotes its unit outward normal vector. More generally, we handle equations driven by monotone operators of $p$-Laplacian type jointly with nonlinear boundary conditions. We prove existence of an entropy solution and check that this solution is unique under natural assumptions. Among other features, we study the regularizing effect given to the solution by both the absorption and the nonlinear boundary term.

Data is a critical element in any discovery process. In the last decades, we observed exponential growth in the volume of available data and the technology to manipulate it. However, data is only practical when one can structure it for a well-defined task. For instance, we need a corpus of text broken into sentences to train a natural language machine-learning model. In this work, we will use the token \textit{dataset} to designate a structured set of data built to perform a well-defined task. Moreover, the dataset will be used in most cases as a blueprint of an entity that at any moment can be stored as a table. Specifically, in science, each area has unique forms to organize, gather and handle its datasets. We believe that datasets must be a first-class entity in any knowledge-intensive process, and all workflows should have exceptional attention to datasets' lifecycle, from their gathering to uses and evolution. We advocate that science and engineering discovery processes are extreme instances of the need for such organization on datasets, claiming for new approaches and tooling. Furthermore, these requirements are more evident when the discovery workflow uses artificial intelligence methods to empower the subject-matter expert. In this work, we discuss an approach to bringing datasets as a critical entity in the discovery process in science. We illustrate some concepts using material discovery as a use case. We chose this domain because it leverages many significant problems that can be generalized to other science fields., Comment: Published at 2nd Annual AAAI Workshop on AI to Accelerate Science and Engineering (AI2ASE) https://ai-2-ase.github.io/papers/16%5cSubmission%5cAAAI_Dataset_Engineering-8.pdf

Evaluating the potential of a prospective candidate is a common task in multiple decision-making processes in different industries. We refer to a prospect as something or someone that could potentially produce positive results in a given context, e.g., an area where an oil company could find oil, a compound that, when synthesized, results in a material with required properties, and so on. In many contexts, assessing the Probability of Success (PoS) of prospects heavily depends on experts' knowledge, often leading to biased and inconsistent assessments. We have developed the framework named KARA (Knowledge-augmented Risk Assessment) to address these issues. It combines multiple AI techniques that consider SMEs (Subject Matter Experts) feedback on top of a structured domain knowledge-base to support risk assessment processes of prospect candidates in knowledge-intensive contexts., Comment: Published at 2nd Annual AAAI Workshop on AI to Accelerate Science and Engineering (AI2ASE) https://ai-2-ase.github.io/papers/17%5cSubmission%5cKaRA_AI2ASE_Workshop-3.pdf. arXiv admin note: text overlap with arXiv:2211.04257

We present observations from X-ray to mid-infrared wavelengths of the most energetic non-quasar transient ever observed, AT2021lwx. Our data show a single optical brightening by a factor $>100$ to a luminosity of $7\times10^{45}$ erg s$^{-1}$, and a total radiated energy of $1.5\times10^{53}$ erg, both greater than any known optical transient. The decline is smooth and exponential and the ultra-violet - optical spectral energy distribution resembles a black body with temperature $1.2\times10^4$ K. Tentative X-ray detections indicate a secondary mode of emission, while a delayed mid-infrared flare points to the presence of dust surrounding the transient. The spectra are similar to recently discovered optical flares in known active galactic nuclei but lack some characteristic features. The lack of emission for the previous seven years is inconsistent with the short-term, stochastic variability observed in quasars, while the extreme luminosity and long timescale of the transient disfavour the disruption of a single solar-mass star. The luminosity could be generated by the disruption of a much more massive star, but the likelihood of such an event occurring is small. A plausible scenario is the accretion of a giant molecular cloud by a dormant black hole of $10^8 - 10^9$ solar masses. AT2021lwx thus represents an extreme extension of the known scenarios of black hole accretion., Comment: 11 pages, 5 figures, Accepted for publication in MNRAS

Accretion disks around compact objects are expected to enter an unstable phase at high luminosity. One instability may occur when the radiation pressure generated by accretion modifies the disk viscosity, resulting in the cyclic depletion and refilling of the inner disk on short timescales. Such a scenario, however, has only been quantitatively verified for a single stellar-mass black hole. Although there are hints of these cycles in a few isolated cases, their apparent absence in the variable emission of most bright accreting neutron stars and black holes has been a lingering puzzle. Here we report the presence of the same multiwavelength instability around an accreting neutron star. Moreover, we show that the variability across the electromagnetic spectrum-from radio to X-ray-of both black holes and neutron stars at high accretion rates can be explained consistently if the accretion disks are unstable, producing relativistic ejections during transitions that deplete or refill the inner disk. Such new association allows us to identify the main physical components responsible for the fast multiwavelength variability of highly accreting compact objects., Comment: Published in Nature. 26 pages, 10 figures. DOI: 10.1038/s41586-022-05648-3

This paper describes our participation in SemEval-2023 Task 10, whose goal is the detection of sexism in social media. We explore some of the most popular transformer models such as BERT, DistilBERT, RoBERTa, and XLNet. We also study different data augmentation techniques to increase the training dataset. During the development phase, our best results were obtained by using RoBERTa and data augmentation for tasks B and C. However, the use of synthetic data does not improve the results for task C. We participated in the three subtasks. Our approach still has much room for improvement, especially in the two fine-grained classifications. All our code is available in the repository https://github.com/isegura/hulat_edos., Comment: The experiments are not reproducible because I did not use a seed for replicability

This paper describes our participation in SemEval-2023 Task 9, Intimacy Analysis of Multilingual Tweets. We fine-tune some of the most popular transformer models with the training dataset and synthetic data generated by different data augmentation techniques. During the development phase, our best results were obtained by using XLM-T. Data augmentation techniques provide a very slight improvement in the results. Our system ranked in the 27th position out of the 45 participating systems. Despite its modest results, our system shows promising results in languages such as Portuguese, English, and Dutch. All our code is available in the repository \url{https://github.com/isegura/hulat_intimacy}.

Assuming a logarithmic interaction between constituent particles, compact and locally accurate wave functions that describe bound states of the two-particle neutral and three-particle charged complexes in two dimensions are designed. Prime examples of these complexes are excitons and trions that appear in monolayers of Transition-Metal DichalCogenides (TMDCs). In the case of excitons, these wave functions led to 5-6 correct decimal digits in the energy and the diamagnetic shifts. In addition, it is demonstrated that they can be used as zero-order approximations to study magnetoexcitons via perturbation theory in powers of the magnetic field strength. For the trion, making a comparison with experimental data for concrete TMDCs, we established that the logarithmic potential leads to binding energies $\lesssim 25\%$ greater than experimental ones. Finally, the structure of the wave function at small distances is established for excitons whose carriers interact via the Rytova-Keldysh potential., Comment: 21 pages, 4 figures

Many real-time applications (e.g., Augmented/Virtual Reality, cognitive assistance) rely on Deep Neural Networks (DNNs) to process inference tasks. Edge computing is considered a key infrastructure to deploy such applications, as moving computation close to the data sources enables us to meet stringent latency and throughput requirements. However, the constrained nature of edge networks poses several additional challenges to the management of inference workloads: edge clusters can not provide unlimited processing power to DNN models, and often a trade-off between network and processing time should be considered when it comes to end-to-end delay requirements. In this paper, we focus on the problem of scheduling inference queries on DNN models in edge networks at short timescales (i.e., few milliseconds). By means of simulations, we analyze several policies in the realistic network settings and workloads of a large ISP, highlighting the need for a dynamic scheduling policy that can adapt to network conditions and workloads. We therefore design ASET, a Reinforcement Learning based scheduling algorithm able to adapt its decisions according to the system conditions. Our results show that ASET effectively provides the best performance compared to static policies when scheduling over a distributed pool of edge resources.

Planetary nebulae (PNe), the ejected envelopes of red giant stars, provide us with a history of the last, mass-losing phases of 90 percent of stars initially more massive than the Sun. Here, we analyse James Webb Space Telescope (JWST) Early Release Observation (ERO) images of the PN NGC3132. A structured, extended H2 halo surrounding an ionised central bubble is imprinted with spiral structures, likely shaped by a low-mass companion orbiting the central star at 40-60 AU. The images also reveal a mid-IR excess at the central star interpreted as a dusty disk, indicative of an interaction with another, closer companion. Including the previously known, A-type visual companion, the progenitor of the NGC3132 PN must have been at least a stellar quartet. The JWST images allow us to generate a model of the illumination, ionisation and hydrodynamics of the molecular halo, demonstrating the power of JWST to investigate complex stellar outflows. Further, new measurements of the A-type visual companion allow us to derive the value for the mass of the progenitor of a central star to date with excellent precision: 2.86+/-0.06 Mo. These results serve as path finders for future JWST observations of PNe providing unique insight into fundamental astrophysical processes including colliding winds, and binary star interactions, with implications for supernovae and gravitational wave systems., Comment: 32 pages, 5 figures for the main article. 12 pages 8 figures for the supplementary material