Search

Efficiently retrieving and synthesizing information from large-scale multimodal collections has become a critical challenge. However, existing video retrieval datasets suffer from scope limitations, primarily focusing on matching descriptive but vague queries with small collections of professionally edited, English-centric videos. To address this gap, we introduce $\textbf{MultiVENT 2.0}$, a large-scale, multilingual event-centric video retrieval benchmark featuring a collection of more than 218,000 news videos and 3,906 queries targeting specific world events. These queries specifically target information found in the visual content, audio, embedded text, and text metadata of the videos, requiring systems leverage all these sources to succeed at the task. Preliminary results show that state-of-the-art vision-language models struggle significantly with this task, and while alternative approaches show promise, they are still insufficient to adequately address this problem. These findings underscore the need for more robust multimodal retrieval systems, as effective video retrieval is a crucial step towards multimodal content understanding and generation tasks.

How are we able to learn about complex current events just from short snippets of video? While natural language enables straightforward ways to represent under-specified, partially observable events, visual data does not facilitate analogous methods and, consequently, introduces unique challenges in event understanding. With the growing prevalence of vision-capable AI agents, these systems must be able to model events from collections of unstructured video data. To tackle robust event modeling in multimodal settings, we introduce a multimodal formulation for partially-defined events and cast the extraction of these events as a three-stage span retrieval task. We propose a corresponding benchmark for this task, MultiVENT-G, that consists of 14.5 hours of densely annotated current event videos and 1,168 text documents, containing 22.8K labeled event-centric entities. We propose a collection of LLM-driven approaches to the task of multimodal event analysis, and evaluate them on MultiVENT-G. Results illustrate the challenges that abstract event understanding poses and demonstrates promise in event-centric video-language systems., Comment: Preprint; 9 pages; 2024 EMNLP Findings

In the present paper, we construct a $\mathbb{Z}/p$-equivariant analog of the $\mathbb{Z}/2$-equivariant spectrum $BP\mathbb{R}$ previously constructed by Hu and Kriz. We prove that this spectrum has some of the properties conjectured by Hill, Hopkins, and Ravenel. Our main construction method is an $\mathbb{Z}/p$-equivariant analog of the Brown-Peterson tower of $BP$, based on a previous description of the $\mathbb{Z}/p$-equivariant Steenrod algebra with constant coefficients by the authors. We also describe several variants of our construction and comparisons with other known equivariant spectra.

In this paper, we study parameter identification for solutions to (possibly non-linear) SDEs driven by additive Rosenblatt process and singularity of the induced laws on the path space. We propose a joint estimator for the drift parameter, diffusion intensity, and Hurst index that can be computed from discrete-time observations with a bounded time horizon and we prove its strong consistency (as well as the speed of convergence) under in-fill asymptotics with a fixed time horizon. As a consequence of this strong consistency, singularity of measures generated by the solutions with different drifts is shown. This results in the invalidity of a Girsanov-type theorem for Rosenblatt processes., Comment: 22 pages, 6 figures

We define new objects called $horizontal$ $p$-$adic$ $L$-$functions$ associated to $L$-values of twists of elliptic curves over $\mathbb{Q}$ by characters of $p$-power order and conductor prime to $p$. We study the fundamental properties of these objects and obtain applications to non-vanishing of finite order twists of central $L$-values, making progress toward conjectures of Goldfeld and David-Fearnley-Kisilevsky. For general elliptic curves $E$ over $\mathbb{Q}$ we obtain strong quantitative lower bounds on the number of non-vanishing central $L$-values of twists by Dirichlet characters of fixed order $d\equiv 2 \text{ mod } 4$ greater than two. We also obtain non-vanishing results for general $d$, including $d = 2$, under mild assumptions. In particular, for elliptic curves with $E[2](\mathbb{Q}) = 0$ we improve on the previously best known lower bounds on the number of non-vanishing $L$-values of quadratic twists due to Ono. Finally, we obtain results on simultaneous non-vanishing of twists of an arbitrary number of elliptic curves with applications to Diophantine stability, as well as results on $p$-adic valuations of moments of $L$-functions., Comment: 52 pages

In this work, we have studied the quasinormal mode, greybody factors, and absorption cross section of de Sitter Reissner-Nordstr\"om black hole surrounded by quintessence field in Rastall gravity. The violation of energy-momentum conservation has a non-linear effect on the quasinormal modes. With an increase in the black hole charge, both real parts of quasinormal modes i.e. oscillation frequency of ring-down Gravitational Waves (GWs) and damping or decay rate of GWs increase non-linearly. A similar observation is made for the black hole structural parameter also, however in this case the variation is almost linear. The remnant mass of a black hole depends on different physical parameters of a black hole. In the case of greybody factors also, we observed that both parameters have similar impacts. With an increase in these parameters, greybody factors decrease. Moreover, the null geodesics and the impact of Rastall gravity on the light trajectory are also investigated. Our study suggests that the presence of a surrounding quintessence field may shadow the existence of black hole charges in such black hole configurations., Comment: 17 pages, 12 figures

Everyday news coverage has shifted from traditional broadcasts towards a wide range of presentation formats such as first-hand, unedited video footage. Datasets that reflect the diverse array of multimodal, multilingual news sources available online could be used to teach models to benefit from this shift, but existing news video datasets focus on traditional news broadcasts produced for English-speaking audiences. We address this limitation by constructing MultiVENT, a dataset of multilingual, event-centric videos grounded in text documents across five target languages. MultiVENT includes both news broadcast videos and non-professional event footage, which we use to analyze the state of online news videos and how they can be leveraged to build robust, factually accurate models. Finally, we provide a model for complex, multilingual video retrieval to serve as a baseline for information retrieval using MultiVENT.

This work explores various manifestations of bumblebee gravity within the metric-affine formalism. We investigate the impact of the Lorentz violation parameter, denoted as $X$, on the modification of the Hawking temperature. Our calculations reveal that as $X$ increases, the values of the Hawking temperature attenuate. To examine the behavior of massless scalar perturbations, specifically the quasinormal modes, we employ the WKB method. The transmission and reflection coefficients are determined through our calculations. The outcomes indicate that a stronger Lorentz-violating parameter results in slower damping oscillations of gravitational waves. To comprehend the influence of the quasinormal spectrum on time-dependent scattering phenomena, we present a detailed analysis of scalar perturbations in the time-domain solution. Additionally, we conduct an investigation on shadows, revealing that larger values of $X$ correspond to larger shadow radii. Furthermore, we constrain the magnitude of the shadow radii using the EHT horizon-scale image of $Sgr A^*$. Finally, we calculate both the time delay and the deflection angle., Comment: 32 pages, 8 figures and 4 tables

As information extraction (IE) systems have grown more adept at processing whole documents, the classic task of template filling has seen renewed interest as benchmark for document-level IE. In this position paper, we call into question the suitability of template filling for this purpose. We argue that the task demands definitive answers to thorny questions of event individuation -- the problem of distinguishing distinct events -- about which even human experts disagree. Through an annotation study and error analysis, we show that this raises concerns about the usefulness of template filling metrics, the quality of datasets for the task, and the ability of models to learn it. Finally, we consider possible solutions., Comment: EMNLP: Findings 2023

In this work, we study the thermodynamic properties on a non--commutative background via gravitational gauge field potentials. This procedure is accomplished after contracting de Sitter (dS) group, $\mathrm{SO}(4,1)$, with the Poincar\`e group, $\mathrm{ISO}(3,1)$. Particularly, we focus on a static spherically symmetric black hole. In this manner, we calculate the modified Hawking temperature and the other deformed thermal state quantities, namely, entropy, heat capacity, Helmholtz free energy and pressure. Finally, we also investigate the black hole evaporation process in such a context., Comment: 14 pages

We identify the equivariant structure of the filtered pieces of the motivic filtration defined by Bhatt, Morrow and Scholze on the topological Hochschild cohomology spectrum of polynomial algebras over $\mathbb{F}_p$., Comment: accepted for publication in the Advances in Mathematics

Contemporary vision benchmarks predominantly consider tasks on which humans can achieve near-perfect performance. However, humans are frequently presented with visual data that they cannot classify with 100% certainty, and models trained on standard vision benchmarks achieve low performance when evaluated on this data. To address this issue, we introduce a procedure for creating datasets of ambiguous images and use it to produce SQUID-E ("Squidy"), a collection of noisy images extracted from videos. All images are annotated with ground truth values and a test set is annotated with human uncertainty judgments. We use this dataset to characterize human uncertainty in vision tasks and evaluate existing visual event classification models. Experimental results suggest that existing vision models are not sufficiently equipped to provide meaningful outputs for ambiguous images and that datasets of this nature can be used to assess and improve such models through model training and direct evaluation of model calibration. These findings motivate large-scale ambiguous dataset creation and further research focusing on noisy visual data., Comment: 10 pages, NeurIPS 2022 Datasets and Benchmarks Track

In this paper, we examine the harmonic oscillator problem in non-commutative phase space (NCPS) by using the Dunkl derivative instead of the habitual one. After defining the Hamilton operator, we use the polar coordinates to derive the binding energy eigenvalue. We find eigenfunctions that correspond to these eigenvalues in terms of the Laguerre functions. We observe that the Dunkl-Harmonic Oscillator (DHO) in the NCPS differs from the ordinary one in the context of providing additional information on the even and odd parities. Therefore, we conclude that working with the Dunkl operator could be more appropriate because of its rich content., Comment: 11 pages 4 Figures

Recent studies have shown that the use of Dunkl derivatives instead of ordinary derivatives leads to deriving parity-dependent dynamic solutions. According to this motivation in this manuscript, we formulate the Dunkl-Schr\"odinger equation within the position-dependent mass formalism and derive an algebraic solution out of it. Our systematic approach lets us observe some new findings in addition to the earlier ones. For example, we find that the solution of the Dunkl-Schr\"oodinger equation with position-dependent mass cannot be considered independent of the choice of parameters. Similarly, through the sl(2) algebra, the energy spectrum and the corresponding wave functions are derived in terms of possible Dunkl, ($\mu$), and mass, ($\alpha$), parameters., Comment: 7 pages

Dunkl derivative enriches solutions by discussing parity due to its reflection operator. Very recently, one of the authors of this manuscript presented one of the most general forms of Dunkl derivative that depends on three Wigner parameters to have a better tuning. In this manuscript, we employ the latter generalized Dunkl derivative in a relativistic equation to examine two-dimensional harmonic and anharmonic oscillators solutions. We obtain the solutions by Nikiforov-Uvarov and QES methods, respectively. We show that degenerate states can occur according to the Wigner parameter values., Comment: 12 pages, 4 figures

Evaluation in machine learning is usually informed by past choices, for example which datasets or metrics to use. This standardization enables the comparison on equal footing using leaderboards, but the evaluation choices become sub-optimal as better alternatives arise. This problem is especially pertinent in natural language generation which requires ever-improving suites of datasets, metrics, and human evaluation to make definitive claims. To make following best model evaluation practices easier, we introduce GEMv2. The new version of the Generation, Evaluation, and Metrics Benchmark introduces a modular infrastructure for dataset, model, and metric developers to benefit from each others work. GEMv2 supports 40 documented datasets in 51 languages. Models for all datasets can be evaluated online and our interactive data card creation and rendering tools make it easier to add new datasets to the living benchmark.

We completely calculate the $RO(\mathbb{Z}/p)$-graded coefficients $H\underline{\mathbb{Z}/p}_\star H\underline{\mathbb{Z}/p}$ for the constant Mackey functor $\underline{\mathbb{Z}/p}$., Comment: Some corrections made and explanations added

The purpose of this paper is mainly to record how certain homotopy-theoretical constructions on ordinary G-equivariant cohomology spectra HM for a Mackey functor M, in particular products and duality, can be described on chain level. We will also discuss certain facts about modules over the constant Green functor $\underline{\mathbb{Z}}$.

We develop a general approach to study three-dimensional Schroedinger operators with confining potentials depending on the distance to a surface. The main idea is to apply parallel coordinates based on the surface but outside its cut locus in the Euclidean space. If the surface is asymptotically planar in a suitable sense, we give an estimate on the location of the essential spectrum of the Schroedinger operator. Moreover, if the surface coincides up to a compact subset with a surface of revolution with strictly positive total Gauss curvature, it is shown that the Schroedinger operator possesses an infinite number of discrete eigenvalues., Comment: proof of Theorem 2 corrected, figures and a conjecture added; to appear in Publ. RIMS, Kyoto University

We consider a general second order linear elliptic equation in a finely perforated domain. The shapes of cavities and their distribution in the domain are arbitrary and non-periodic; they are supposed to satisfy minimal natural geometric conditions. On the boundaries of the cavities we impose either the Dirichlet or a nonlinear Robin condition; the choice of the type of the boundary condition for each cavity is arbitrary. Then we suppose that for some cavities the nonlinear Robin condition is sign-definite in certain sense. Provided such cavities and ones with the Dirichlet condition are distributed rather densely in the domain and the characteristic sizes of the cavities and the minimal distances between the cavities satisfy certain simple condition, we show that a solution to our problem tends to zero as the perforation becomes finer. Our main result are order sharp estimates for the $L_2$- and $W_2^1$-norms of the solution uniform in the $L_2$-norm of the right hand side in the equation.

While popular televised events such as presidential debates or TV shows are airing, people provide commentary on them in real-time. In this paper, we propose a simple yet effective approach to combine social media commentary and videos to create a multimedia summary of televised events. Our approach identifies scenes from these events based on spikes of mentions of people involved in the event and automatically selects tweets and frames from the videos that occur during the time period of the spike that talk about and show the people being discussed., Comment: 8 pages, 3 figures, 7 tables

This paper deals with the drift estimation in linear stochastic evolution equations (with emphasis on linear SPDEs) with additive fractional noise (with Hurst index ranging from 0 to 1) via least-squares procedure. Since the least-squares estimator contains stochastic integrals of divergence type, we address the problem of its pathwise (and robust to observation errors) evaluation by comparison with the pathwise integral of Stratonovich type and using its chain-rule property. The resulting pathwise LSE is then defined implicitly as a solution to a non-linear equation. We study its numerical properties (existence and uniqueness of the solution) as well as statistical properties (strong consistency and the speed of its convergence). The asymptotic properties are obtained assuming fixed time horizon and increasing number of the observed Fourier modes (space asymptotics). We also conjecture the asymptotic normality of the pathwise LSE., Comment: 30 pages, 6 figures

In this manuscript, the modification of thermodynamic properties of Schwarzschild and Reissner-Nordstr\"{o}m black holes in the presence of higher-order generalized uncertainty principle (GUP) have been investigated. We considered heuristic analysis versus the behavior of a particle that is absorbed by the black hole, and then, we studied the thermodynamic properties of the black holes such as temperature-mass, temperature-charge-mass, entropy, and heat capacity functions up to the first and second-order of the GUP parameter. We compared the behaviors of the GUP-corrected thermodynamic properties for both black holes with graphical methods., Comment: 15 pages, 6 figures

We report on the first Belle II measurement of the branching fraction ($\mathcal{B}$) and longitudinal polarization fraction ($f_L$) of $B^+\to \rho^+\rho^0$ decays. We reconstruct $B^+\to \rho^+(\to \pi^+\pi^0(\to \gamma\gamma))\rho^0(\to \pi^+\pi^-)$ decays in a sample of SuperKEKB electron-positron collisions collected by the Belle II experiment in 2019 and 2020 at the $\Upsilon$(4S) resonance and corresponding to $62.8$ fb$^{-1}$ of integrated luminosity. We fit the distributions of the difference between expected and observed $B$ candidate energy, continuum-suppression variable, dipion masses, and angular distributions of the resulting samples, to determine a signal yield of $104\pm16$ events. The signal yields are corrected for efficiencies determined from simulation and control data samples to obtain $\mathcal{B}(B^+ \to \rho^+\rho^0) = [20.6 \pm 3.2(\rm stat) \pm 4.0(\rm syst)]\times 10^{-6}$, and $f_L(B^+ \to \rho^+\rho^0) = 0.936 ^{+0.049}_{-0.041}(\rm stat)\pm 0.021(\rm syst)$. This first Belle II $B^+ \to \rho^+\rho^0$ angular analysis yields results compatible with previous determinations, and indicates Belle II performance superior to early Belle results.

We report on measurements of branching fractions ($\mathcal{B}$) and CP-violating charge asymmetries ($\mathcal{A}_{\rm CP}$) of multibody charmless $B$ decays reconstructed by the Belle II experiment at the SuperKEKB electron-positron collider. We use a sample of collisions collected in 2019 and 2020 at the $\Upsilon(4S)$ resonance and corresponding to $62.8$ fb$^{-1}$ of integrated luminosity. We use simulation to determine optimized event selections. The $\Delta E$ and $M_{\rm bc}$ distributions of the resulting samples are fit to determine signal yields of approximately 690, 840, and 380 decays for the channels $B^+ \to K^+K^-K^+$, $B^+ \to K^+\pi^-\pi^+$, and $B^0 \to K^+\pi^-\pi^0$, respectively. These yields are corrected for efficiencies determined from simulation and control data samples to obtain $\mathcal{B}(B^+ \to K^+K^-K^+) = [35.8 \pm 1.6(\rm stat) \pm 1.4 (\rm syst)]\times 10^{-6}$, $\mathcal{B}(B^+ \to K^+\pi^-\pi^+) = [67.0 \pm 3.3 (\rm stat)\pm 2.3 (\rm syst)]\times 10^{-6}$, $\mathcal{B}(B^0 \to K^+\pi^-\pi^0) = [38.1 \pm 3.5 (\rm stat)\pm 3.9 (\rm syst)]\times 10^{-6}$, $\mathcal{A}_{\rm CP}(B^+ \to K^+K^-K^+) = -0.103 \pm 0.042(\rm stat) \pm 0.020 (\rm syst)$, $\mathcal{A}_{\rm CP}(B^+ \to K^+\pi^-\pi^+) = -0.010 \pm 0.050 (\rm stat)\pm 0.021(\rm syst)$, and $\mathcal{A}_{\rm CP}(B^0 \to K^+\pi^-\pi^0) = 0.207 \pm 0.088 (\rm stat)\pm 0.011(\rm syst)$. Results are consistent with previous measurements and demonstrate detector performance comparable with the best Belle results.

An important task in NLP applications such as sentence simplification is the ability to take a long, complex sentence and split it into shorter sentences, rephrasing as necessary. We introduce a novel dataset and a new model for this `split and rephrase' task. Our BiSECT training data consists of 1 million long English sentences paired with shorter, meaning-equivalent English sentences. We obtain these by extracting 1-2 sentence alignments in bilingual parallel corpora and then using machine translation to convert both sides of the corpus into the same language. BiSECT contains higher quality training examples than previous Split and Rephrase corpora, with sentence splits that require more significant modifications. We categorize examples in our corpus, and use these categories in a novel model that allows us to target specific regions of the input sentence to be split and edited. Moreover, we show that models trained on BiSECT can perform a wider variety of split operations and improve upon previous state-of-the-art approaches in automatic and human evaluations., Comment: 9 pages, 9 figures. Long paper to appear in Empirical Methods in Natural Language Processing 2021 (EMNLP 2021)

In this paper, we calculate the coefficient ring of equivariant Thom complex cobordism for the symmetric group on three elements. We also make some remarks on general methods of calculating certain pullbacks of rings which typically occur in calculations of equivariant cobordism., Comment: This article was first written in November 2018. The present version contains some minor corrections