Search

Disinformation on social media poses both societal and technical challenges. While previous studies have integrated textual information into propagation networks, they have yet to fully leverage the advancements in Transformer-based language models for high-quality contextual text representations. This work investigates the impact of incorporating textual features into Graph Neural Networks (GNNs) for fake news detection. Our experiments demonstrate that contextual representations improve performance by 9.3% in Macro F1 over static ones and 33.8% over GNNs without textual features. However, noisy data augmentation degrades performance and increases instability. We expect our methodology to open avenues for further research, and all code is made publicly available., Comment: Work in progress

In a context where the Brazilian judiciary system, the largest in the world, faces a crisis due to the slow processing of millions of cases, it becomes imperative to develop efficient methods for analyzing legal texts. We introduce uBERT, a hybrid model that combines Transformer and Recurrent Neural Network architectures to effectively handle long legal texts. Our approach processes the full text regardless of its length while maintaining reasonable computational overhead. Our experiments demonstrate that uBERT achieves superior performance compared to BERT+LSTM when overlapping input is used and is significantly faster than ULMFiT for processing long legal documents., Comment: To appear at 15th STIL @ BRACIS'24

Understanding trajectories in multi-agent scenarios requires addressing various tasks, including predicting future movements, imputing missing observations, inferring the status of unseen agents, and classifying different global states. Traditional data-driven approaches often handle these tasks separately with specialized models. We introduce TranSPORTmer, a unified transformer-based framework capable of addressing all these tasks, showcasing its application to the intricate dynamics of multi-agent sports scenarios like soccer and basketball. Using Set Attention Blocks, TranSPORTmer effectively captures temporal dynamics and social interactions in an equivariant manner. The model's tasks are guided by an input mask that conceals missing or yet-to-be-predicted observations. Additionally, we introduce a CLS extra agent to classify states along soccer trajectories, including passes, possessions, uncontrolled states, and out-of-play intervals, contributing to an enhancement in modeling trajectories. Evaluations on soccer and basketball datasets show that TranSPORTmer outperforms state-of-the-art task-specific models in player forecasting, player forecasting-imputation, ball inference, and ball imputation. https://youtu.be/8VtSRm8oGoE, Comment: Accepted to ACCV 2024

We propose an inflation scenario with three independent stages of cold, warm and thermal inflation, respectively, driven by different scalar fields, motivated by the large number of such fields predicted in most extensions of the Standard Model. We show, in particular, that the intermediate period of warm inflation naturally leads to large density fluctuations on small scales, which can lead to primordial black hole formation in the mass window where they may account for all dark matter. This type of scenario yields a distinctive primordial black hole mass function with a mass gap, with the final period of thermal inflation diluting the abundance of very light black holes., Comment: 26 pages, 5 figures

Instruction following is a key capability for LLMs. However, recent studies have shown that LLMs often struggle with instructions containing multiple constraints (e.g. a request to create a social media post "in a funny tone" with "no hashtag"). Despite this, most evaluations focus solely on synthetic data. To address this, we introduce RealInstruct, the first benchmark designed to evaluate LLMs' ability to follow real-world multi-constrained instructions by leveraging queries real users asked AI assistants. We also investigate model-based evaluation as a cost-effective alternative to human annotation for this task. Our findings reveal that even the proprietary GPT-4 model fails to meet at least one constraint on over 21% of instructions, highlighting the limitations of state-of-the-art models. To address the performance gap between open-source and proprietary models, we propose the Decompose, Critique and Refine (DeCRIM) self-correction pipeline, which enhances LLMs' ability to follow constraints. DeCRIM works by decomposing the original instruction into a list of constraints and using a Critic model to decide when and where the LLM's response needs refinement. Our results show that DeCRIM improves Mistral's performance by 7.3% on RealInstruct and 8.0% on IFEval even with weak feedback. Moreover, we demonstrate that with strong feedback, open-source LLMs with DeCRIM can outperform GPT-4 on both benchmarks., Comment: To appear at EMNLP 2024

In this opinion article, we gathered some reflections and practical tips on what Early Stage Researchers do against inequalities in academia. This is the longer version of an opinion paper that was recently published on the website of the International Society for Optics and Photonics (spie.org), containing more details and suggestions., Comment: comments and feedback are welcome

In this work, we use the local spin Chern marker (LSCM) recently introduced by Ba\`{u} and Marrazzo [Phys. Rev. B 110, 054203 (2024)] to analyze the topology of the ground-state electronic wave functions in a finite honeycomb lattice flake described by three distinct models. The models considered here are characterized by strong Rashba spin-orbit interaction, which leads to non-conservation of the spin operator, i.e., $[\mathcal{H},\hat{s}_z]\neq 0$. The three spin-orbit couplings associated with the topological aspects of the models are: 1) Standard Kane-Mele coupling, 2) Sublattice-dependent Kane-Mele coupling, and 3) In-plane ($\hat{s}_y$) polarized Kane-Mele coupling. These couplings occur in graphene grown on suitable substrates and are relevant for modeling its van der Waals heterostructures. A particular topological phase diagram characterizes each of these spin-orbit interactions, and our calculations of LSCM successfully capture its general features. We also performed a detailed analysis of the spectral properties of the energy and valence-projected spin matrix eigenvalues, which shows that both exhibit a gap that protects the marker. Our results expand the applicability of the spin Chern number method to a class of Hamiltonians with experimental relevance and may contribute to future research on the real-space topology of realistic materials.

It is established ground states and multiplicity of solutions for a nonlocal Schr\"{o}dinger equation $(-\Delta )^s u + V(x) u = \lambda a(x) |u|^{q-2}u + b(x)f(u)$ in $\mathbb{R}^N,$ $u \in H^s(\mathbb{R}^N),$ where $00,$ under general conditions over the measurable functions $a,$ $b$, $V$ and $f.$ The nonlinearity $f$ is superlinear at infinity and at the origin, and does not satisfy any Ambrosetti-Rabinowitz type condition. It is considered that the weights $a$ and $b$ are not necessarily bounded and the potential $V$ can change sign. We obtained a sharp $\lambda^*> 0$ which guarantees the existence of at least two nontrivial solutions for each $\lambda \in (0, \lambda^*)$. Our approach is variational in its nature and is based on the nonlinear Rayleigh quotient method together with some fine estimates. Compactness of the problem is also considered.

We show that topological superconductivity can be produced by strong electron-electron (e-e) interaction in two minimal one (1D) and two-dimensional (2D) models containing only nearest neighbor hopping and suitable superconducting order parameters. The presence of the strong e-e interaction constrains each atomic site to be either empty or singly occupied. One can treat this scenario by fractionalizing the electrons into spin and charge degrees of freedom. We use the coherent state symbols associated with the lowest irreducible representation of the $su(2|1)$ superalgebra spanned by the Hubbard operators to solve the proposed models, as they implicitly take into account both the single particle occupation constraint and the fractionalization of the electrons. Our two example display the emergence of topological superconductivity in an 1D nanowire with extended $s$-wave superconductivity, and in a 2D square lattice with $p$-wave superconductivity., Comment: 9 pages, 2 figures

Motion prediction in soccer involves capturing complex dynamics from player and ball interactions. We present FootBots, an encoder-decoder transformer-based architecture addressing motion prediction and conditioned motion prediction through equivariance properties. FootBots captures temporal and social dynamics using set attention blocks and multi-attention block decoder. Our evaluation utilizes two datasets: a real soccer dataset and a tailored synthetic one. Insights from the synthetic dataset highlight the effectiveness of FootBots' social attention mechanism and the significance of conditioned motion prediction. Empirical results on real soccer data demonstrate that FootBots outperforms baselines in motion prediction and excels in conditioned tasks, such as predicting the players based on the ball position, predicting the offensive (defensive) team based on the ball and the defensive (offensive) team, and predicting the ball position based on all players. Our evaluation connects quantitative and qualitative findings. https://youtu.be/9kaEkfzG3L8, Comment: Published as a conference paper at IEEE ICIP 2024

We present a novel approach to quantum algorithms, by taking advantage of modular values, i.e., complex and unbounded quantities resulting from specific post-selected measurement scenarios. Our focus is on the problem of ascertaining whether a given function acting on a set of binary values is constant (uniformly yielding outputs of either all 0 or all 1), or balanced (a situation wherein half of the outputs are 0 and the other half are 1). Such problem can be solved by relying on the Deutsch-Jozsa algorithm. The proposed method, relying on the use of modular values, provides a high number of degrees of freedom for optimizing the new algorithm inspired from the Deutsch-Jozsa one. In particular, we explore meticulously the choices of the pre- and post-selected states. We eventually test the novel theoretical algorithm on a quantum computing platform. While the outcomes are currently not on par with the conventional approach, they nevertheless shed light on potential for future improvements, especially with less-optimized algorithms. We are thus confidend that the proposed proof of concept could prove its validity in bridging quantum algorithms and modular values research fields.

PLATO (PLAnetary Transits and Oscillations of stars) is ESA's M3 mission designed to detect and characterise extrasolar planets and perform asteroseismic monitoring of a large number of stars. PLATO will detect small planets (down to <2 R_(Earth)) around bright stars (<11 mag), including terrestrial planets in the habitable zone of solar-like stars. With the complement of radial velocity observations from the ground, planets will be characterised for their radius, mass, and age with high accuracy (5 %, 10 %, 10 % for an Earth-Sun combination respectively). PLATO will provide us with a large-scale catalogue of well-characterised small planets up to intermediate orbital periods, relevant for a meaningful comparison to planet formation theories and to better understand planet evolution. It will make possible comparative exoplanetology to place our Solar System planets in a broader context. In parallel, PLATO will study (host) stars using asteroseismology, allowing us to determine the stellar properties with high accuracy, substantially enhancing our knowledge of stellar structure and evolution. The payload instrument consists of 26 cameras with 12cm aperture each. For at least four years, the mission will perform high-precision photometric measurements. Here we review the science objectives, present PLATO's target samples and fields, provide an overview of expected core science performance as well as a description of the instrument and the mission profile at the beginning of the serial production of the flight cameras. PLATO is scheduled for a launch date end 2026. This overview therefore provides a summary of the mission to the community in preparation of the upcoming operational phases.

Whisper is a multitask and multilingual speech model covering 99 languages. It yields commendable automatic speech recognition (ASR) results in a subset of its covered languages, but the model still underperforms on a non-negligible number of under-represented languages, a problem exacerbated in smaller model versions. In this work, we examine its limitations, demonstrating the presence of speaker-related (gender, age) and model-related (resourcefulness and model size) bias. Despite that, we show that only model-related bias are amplified by quantization, impacting more low-resource languages and smaller models. Searching for a better compression approach, we propose DistilWhisper, an approach that is able to bridge the performance gap in ASR for these languages while retaining the advantages of multitask and multilingual capabilities. Our approach involves two key strategies: lightweight modular ASR fine-tuning of whisper-small using language-specific experts, and knowledge distillation from whisper-large-v2. This dual approach allows us to effectively boost ASR performance while keeping the robustness inherited from the multitask and multilingual pre-training. Results demonstrate that our approach is more effective than standard fine-tuning or LoRA adapters, boosting performance in the targeted languages for both in- and out-of-domain test sets, while introducing only a negligible parameter overhead at inference., Comment: Master Thesis

In this paper, we investigate the impact of bond-dilution disorder on the critical behavior of the stochastic SIR model. Monte Carlo simulations were conducted using square lattices with first- and second-nearest neighbor interactions. Quenched bond-diluted lattice disorder was introduced into the systems, allowing them to evolve over time. By employing percolation theory and finite-size scaling analysis, we estimate both the critical threshold and leading critical exponent ratios of the model for different bond-dilution rates ($p$). An examination of the average size of the percolating cluster and the size distribution of non-percolating clusters of recovered individuals was performed to ascertain the universality class of the model. The simulation results strongly indicate that the present model belongs to a new universality class distinct from that of 2D dynamical percolation, depending on the specific $p$ value under consideration., Comment: 10 pages and 10 figures. arXiv admin note: substantial text overlap with arXiv:2110.01054

In this paper we formalize, using the Nyquist-Shannon theorem, a fundamental temporal resolution limit for numerical experiments in complex systems. A consequence of this limit is aliasing, the introduction of spurious frequencies due to sampling. By imposing these limits on the uncertainty principle in harmonic analysis, we show that by increasing the sampling interval $\Delta t$, we can also artificially stretch the temporal behavior of our numerical experiment. Importantly, in limiting cases, we could even observe a new artificially created absorbing state. Our findings are validated in deterministic and stochastic simulations. In deterministic systems, we analyzed the Kuramoto model in which aliasing could be observed. In stochastic simulations, we formalized and compared different simulation approaches and showed their temporal limits. Gillespie-like simulations fully capture the continuous-time Markov chain processes, being lossless. Asynchronous cellular automata methods capture the same transitions as the continuous-time process but lose the temporal information about the process. Finally, synchronous cellular automata simulations solve a sampled chain. By comparing these methods, we show that if $\Delta t$ is not small enough, the cellular automata approach fails to capture the original continuous-time Markov chain since the sampling is already built into the simulation method. Our results point to a fundamental limitation that cannot be overcome by traditional methods of numerical simulations.

We propose a team assignment algorithm based on a hypergraph approach focusing on resilience and diffusion optimization. Specifically, our method is based on optimizing the algebraic connectivity of the Laplacian matrix of an edge-dependent vertex-weighted hypergraph. We used constrained simulated annealing, where we constrained the effort agents can exert to perform a task and the minimum effort a task requires to be completed. We evaluated our methods in terms of the number of unsuccessful patches to drive our solution into the feasible region and the cost of patching. We showed that our formulation provides more robust solutions than the original data and the greedy approach. We hope that our methods motivate further research in applying hypergraphs to similar problems in different research areas and in exploring variations of our methods., Comment: 18 pages, 12 Figures

We analyse Hamiltonian-type systems of second-order elliptic PDE invariant under a non-compact group and, consequently, involve a lack of compactness of the Sobolev embedding. We show that the loss of compactness can be compensated by using a concentration-compactness principle via weak profile decomposition for bounded Palais-Smale sequences in Banach spaces. Our analysis to prove the existence of ground states involves a reduction by the inversion method of the system to a fourth-order equation combined with a variational principle of a minimax nature. Among other results, including regularity and a Pohozaev-type identity, we also prove the non-existence of weak solutions for a class of Lane-Emden systems.

Understanding the dissemination of diseases, information, and behavior stands as a paramount research challenge in contemporary network and complex systems science. The COVID-19 pandemic and the proliferation of misinformation are relevant examples of the importance of these dynamic processes, which have recently gained more attention due to the potential of higher-order networks to unlock new avenues for their investigation. Despite being in its early stages, the examination of social contagion in higher-order networks has witnessed a surge of novel research and concepts, revealing different functional forms for the spreading dynamics and offering novel insights. This review presents a focused overview of this body of literature and proposes a unified formalism that covers most of these forms. The goal is to underscore the similarities and distinctions among various models, to motivate further research on the general and universal properties of such models. We also highlight that while the path for additional theoretical exploration appears clear, the empirical validation of these models through data or experiments remains scant, with an unsettled roadmap as of today. We therefore conclude with some perspectives aimed at providing possible research directions that could contribute to a better understanding of this class of dynamical processes, both from a theoretical and a data-oriented point of view., Comment: Review article. 17 pages and 5 figures. Submitted for publication

Districting-and-routing is a strategic problem aiming to aggregate basic geographical units (e.g., zip codes) into delivery districts. Its goal is to minimize the expected long-term routing cost of performing deliveries in each district separately. Solving this stochastic problem poses critical challenges since repeatedly evaluating routing costs on a set of scenarios while searching for optimal districts takes considerable time. Consequently, solution approaches usually replace the true cost estimation with continuous cost approximation formulas extending Beardwood-Halton-Hammersley and Daganzo's work. These formulas commit errors that can be magnified during the optimization step. To reconcile speed and solution quality, we introduce a supervised learning and optimization methodology leveraging a graph neural network for delivery-cost estimation. This network is trained to imitate known costs generated on a limited subset of training districts. It is used within an iterated local search procedure to produce high-quality districting plans. Our computational experiments, conducted on five metropolitan areas in the United Kingdom, demonstrate that the graph neural network predicts long-term district cost operations more accurately, and that optimizing over this oracle permits large economic gains (10.12% on average) over baseline methods that use continuous approximation formulas or shallow neural networks. Finally, we observe that having compact districts alone does not guarantee high-quality solutions and that other learnable geometrical features of the districts play an essential role.

Artificial intelligence (AI) has driven many information and communication technology (ICT) breakthroughs. Nonetheless, the scope of ICT systems has expanded far beyond AI since the Turing test proposal. Critically, recent AI regulation proposals adopt AI definitions affecting ICT techniques, approaches, and systems that are not AI. In some cases, even works from mathematics, statistics, and engineering would be affected. Worryingly, AI misdefinitions are observed from Western societies to the Global South. In this paper, we propose a framework to score how validated as appropriately-defined for regulation (VADER) an AI definition is. Our online, publicly-available VADER framework scores the coverage of premises that should underlie AI definitions for regulation, which aim to (i) reproduce principles observed in other successful technology regulations, and (ii) include all AI techniques and approaches while excluding non-AI works. Regarding the latter, our score is based on a dataset of representative AI, non-AI ICT, and non-ICT examples. We demonstrate our contribution by reviewing the AI regulation proposals of key players, namely the United States, United Kingdom, European Union, and Brazil. Importantly, none of the proposals assessed achieve the appropriateness score, ranging from a revision need to a concrete risk to ICT systems and works from other fields.

A major challenge in Natural Language Processing is obtaining annotated data for supervised learning. An option is the use of crowdsourcing platforms for data annotation. However, crowdsourcing introduces issues related to the annotator's experience, consistency, and biases. An alternative is to use zero-shot methods, which in turn have limitations compared to their few-shot or fully supervised counterparts. Recent advancements driven by large language models show potential, but struggle to adapt to specialized domains with severely limited data. The most common approaches therefore involve the human itself randomly annotating a set of datapoints to build initial datasets. But randomly sampling data to be annotated is often inefficient as it ignores the characteristics of the data and the specific needs of the model. The situation worsens when working with imbalanced datasets, as random sampling tends to heavily bias towards the majority classes, leading to excessive annotated data. To address these issues, this paper contributes an automatic and informed data selection architecture to build a small dataset for few-shot learning. Our proposal minimizes the quantity and maximizes diversity of data selected for human annotation, while improving model performance., Comment: Accepted at PROPOR 2024 - The 16th International Conference on Computational Processing of Portuguese

Building on recently constructed inflationary vector dark matter production mechanisms as well as studies of magnetogenesis, we show that an inflationary dark Schwinger mechanism can generate the observed dark matter relic abundance for `dark electron' masses as light as $\sim 0.1$ eV and as heavy as $10^{12}$ GeV. The dark matter can interact very weakly via the exchange of light dark photons with a power spectrum which is peaked at very small scales, thus evading isocurvature constraints. This mechanism is viable even when (purely) gravitational particle production is negligible. Thus dark matter can be produced solely via the Schwinger effect during inflation including for light masses., Comment: Major updates, new constraints derived, 2 Figures

In this review article, we revisit the topic of baryogenesis, which is the physical process that generated the observed baryon asymmetry during the first stages of the primordial Universe. A viable theoretical explanation to understand and investigate the mechanisms underlying baryogenesis must always ensure that the Sakharov criteria are fulfilled. These essentially state the following: (i) baryon number violation; (ii) the violation of both C (charge conjugation symmetry) and CP (the composition of parity and C); (iii) and the departure from equilibrium. Throughout the years, various mechanisms have been proposed to address this issue, and here we review two of the most important, namely, electroweak baryogenesis (EWB) and Grand Unification Theories (GUTs) baryogenesis. Furthermore, we briefly explore how a change in the theory of gravity affects the EWB and GUT baryogenesis by considering Scalar--Tensor Theories (STT), where the inclusion of a scalar field mediates the gravitational interaction, in addition to the metric tensor field. We consider specific STT toy models and show that a modification of the underlying gravitational theory implies a change in the time--temperature relation of the evolving cosmological model, thus altering the conditions that govern the interplay between the rates of the interactions generating baryon asymmetry, and the expansion rate of the Universe. Therefore, the equilibrium of the former does not exactly occur as in the general relativistic standard model, and there are consequences for the baryogenesis mechanisms that have been devised. This is representative of the type of modifications of the baryogenesis processes that are to be found when considering extended theories of gravity., Comment: 29 pages. Published version in Symmetry of an invited contribution to the Special Issue "Physics and Symmetry Section: Feature Papers 2023"