Search

Self-supervised learning (SSL) is used in deep learning to train on large datasets without the need for expensive labelling of the data. Recently, large Automatic Speech Recognition (ASR) models such as XLS-R have utilised SSL to train on over one hundred different languages simultaneously. However, deeper investigation shows that the bulk of the training data for XLS-R comes from a small number of languages. Biases learned through SSL have been shown to exist in multiple domains, but language bias in multilingual SSL ASR has not been thoroughly examined. In this paper, we utilise the Lottery Ticket Hypothesis (LTH) to identify language-specific subnetworks within XLS-R and test the performance of these subnetworks on a variety of different languages. We are able to show that when fine-tuning, XLS-R bypasses traditional linguistic knowledge and builds only on weights learned from the languages with the largest data contribution to the pretraining data., Comment: Accepted to Speech and Language Technology Workshop (SLT) 2024 accessible on IEEE Xplore

Sub-Neptunes are the most common type of planet in our galaxy. Interior structure models suggest that the coldest sub-Neptunes could host liquid water oceans underneath their hydrogen envelopes - sometimes called 'hycean' planets. JWST transmission spectra of the $\sim$ 250 K sub-Neptune K2-18 b were recently used to report detections of CH$_4$ and CO$_2$, alongside weaker evidence of (CH$_3$)$_2$S (dimethyl sulfide, or DMS). Atmospheric CO$_2$ was interpreted as evidence for a liquid water ocean, while DMS was highlighted as a potential biomarker. However, these notable claims were derived using a single data reduction and retrieval modeling framework, which did not allow for standard robustness tests. Here we present a comprehensive reanalysis of K2-18 b's JWST NIRISS SOSS and NIRSpec G395H transmission spectra, including the first analysis of the second-order NIRISS SOSS data. We incorporate multiple well-tested data reduction pipelines and retrieval codes, spanning 60 different data treatments and over 250 atmospheric retrievals. We confirm the detection of CH$_4$ ($\approx$ 4$\sigma$), with a volume mixing ratio of log CH$_4$ = $-1.15^{+0.40}_{-0.52}$, but we find no statistically significant or reliable evidence for CO$_2$ or DMS. Finally, we quantify the observed atmospheric composition using photochemical-climate and interior models, demonstrating that our revised composition of K2-18 b can be explained by an oxygen-poor mini-Neptune without requiring a liquid water surface or life., Comment: 42 pages, 20 figures. Submitted to AAS Journals

The increasing complexity of application requirements and the dynamic nature of the Cloud-Edge Continuum present significant challenges for efficient resource management. These challenges stem from the ever-changing infrastructure, which is characterized by additions, removals, and reconfigurations of nodes and links, as well as the variability of application workloads. Traditional centralized approaches struggle to adapt to these changes due to their static nature, while decentralized solutions face challenges such as limited global visibility and coordination overhead. This paper proposes a hybrid decentralized framework for dynamic application placement and resource management. The framework utilizes Graph Neural Networks (GNNs) to embed resource and application states, enabling comprehensive representation and efficient decision-making. It employs a collaborative multi-agent reinforcement learning (MARL) approach, where local agents optimize resource management in their neighborhoods and a global orchestrator ensures system-wide coordination. By combining decentralized application placement with centralized oversight, our framework addresses the scalability, adaptability, and accuracy challenges inherent in the Cloud-Edge Continuum. This work contributes to the development of decentralized application placement strategies, the integration of GNN embeddings, and collaborative MARL systems, providing a foundation for efficient, adaptive and scalable resource management.

Core spanners are a class of document spanners that capture the core functionality of IBM's AQL. FC is a logic on strings built around word equations that when extended with constraints for regular languages can be seen as a logic for core spanners. The recently introduced FC-Datalog extends FC with recursion, which allows us to define recursive relations for core spanners. Additionally, as FC-Datalog captures P, it is also a tractable version of Datalog on strings. This presents an opportunity for optimization. We propose a series of FC-Datalog fragments with desirable properties in terms of complexity of model checking, expressive power, and efficiency of checking membership in the fragment. This leads to a range of fragments that all capture LOGSPACE, which we further restrict to obtain linear combined complexity. This gives us a framework to tailor fragments for particular applications. To showcase this, we simulate deterministic regex in a tailored fragment of FC-Datalog.

In the standard picture of fully-developed turbulence, highly intermittent hydrodynamic fields are nonlinearly coupled across scales, where local energy cascades from large scales into dissipative vortices and large density gradients. Microscopically, however, constituent fluid molecules are in constant thermal (Brownian) motion, but the role of molecular fluctuations on large-scale turbulence is largely unknown, and with rare exceptions, it has historically been considered irrelevant at scales larger than the molecular mean free path. Recent theoretical and computational investigations have shown that molecular fluctuations can impact energy cascade at Kolmogorov length scales. Here we show that molecular fluctuations not only modify energy spectrum at wavelengths larger than the Kolmogorov length in compressible turbulence, but they also significantly inhibit spatio-temporal intermittency across the entire dissipation range. Using large-scale direct numerical simulations of computational fluctuating hydrodynamics, we demonstrate that the extreme intermittency characteristic of turbulence models is replaced by nearly-Gaussian statistics in the dissipation range. These results demonstrate that the compressible Navier-Stokes equations should be augmented with molecular fluctuations to accurately predict turbulence statistics across the dissipation range. Our findings have significant consequences for turbulence modeling in applications such as astrophysics, reactive flows, and hypersonic aerodynamics, where dissipation-range turbulence is approximated by closure models., Comment: 13 pages, 5 figures

The Safe Trusted Autonomy for Responsible Space (STARS) program aims to advance autonomy technologies for space by leveraging machine learning technologies while mitigating barriers to trust, such as uncertainty, opaqueness, brittleness, and inflexibility. This paper presents the achievements and lessons learned from the STARS program in integrating reinforcement learning-based multi-satellite control, run time assurance approaches, and flexible human-autonomy teaming interfaces, into a new integrated testing environment for collaborative autonomous satellite systems. The primary results describe analysis of the reinforcement learning multi-satellite control and run time assurance algorithms. These algorithms are integrated into a prototype human-autonomy interface using best practices from human-autonomy trust literature, however detailed analysis of the effectiveness is left to future work. References are provided with additional detailed results of individual experiments.

We present the Cosmic Evolution Early Release Science (CEERS) Survey, a 77.2 hour Director's Discretionary Early Release Science Program. CEERS demonstrates, tests, and validates efficient extragalactic surveys using coordinated, overlapping parallel observations with the JWST instrument suite, including NIRCam and MIRI imaging, NIRSpec low (R~100) and medium (R~1000) resolution spectroscopy, and NIRCam slitless grism (R~1500) spectroscopy. CEERS targets the Hubble Space Telescope-observed region of the Extended Groth Strip (EGS) field, supported by a rich set of multiwavelength data. CEERS facilitated immediate community science in both of the extragalactic core JWST science drivers ``First Light" and ``Galaxy Assembly," including: 1) The discovery and characterization of large samples of galaxies at z >~ 10 from ~90 arcmin^2 of NIRCam imaging, constraining their abundance and physical nature; 2) Deep spectra of >1000 galaxies, including dozens of galaxies at 63; and 4) Characterizing galaxy mid-IR emission with MIRI to study dust-obscured star-formation and supermassive black hole growth at z~1-3. As a legacy product for the community, the CEERS team has provided several data releases, accompanied by detailed notes on the data reduction procedures and notebooks to aid in reproducibility. In addition to an overview of the survey and quality of the data, we provide science highlights from the first two years with CEERS data., Comment: 38 pages, 13 figures, 6 tables

We report the discovery and characterization of two sub-Saturns from the Transiting Exoplanet Survey Satellite (\textit{TESS}) using high-resolution spectroscopic observations from the MaHPS spectrograph at the Wendelstein Observatory and the SOPHIE spectrograph at the Haute-Provence Observatory. Combining photometry from TESS, KeplerCam, LCOGT, and MuSCAT2 with the radial velocity measurements from MaHPS and SOPHIE we measure precise radii and masses for both planets. TOI-5108 b is a sub-Saturn with a radius of $6.6 \pm 0.1$ $R_\oplus$ and a mass of $32 \pm 5$ $M_\oplus$. TOI-5786 b is similar to Saturn with a radius of $8.54 \pm 0.13$ $R_\oplus$ and a mass of $73 \pm 9$ $M_\oplus$. The host star for TOI-5108 b is a moderately bright (Vmag 9.75) G-type star. TOI-5786 is a slightly dimmer (Vmag 10.2) F-type star. Both planets are close to their host stars with periods of 6.75 days and 12.78 days respectively. This puts TOI-5108 b just inside the bounds of the Neptune desert while TOI-5786 b is right above the upper edge. We estimate hydrogen-helium envelope mass fractions of $38 \%$ for TOI-5108 b and $74 \% $ for TOI-5786 b. However, using a model for the interior structure that includes tidal effects the envelope fraction of TOI-5108 b could be much lower ($\sim 20\,\%$) depending on the obliquity. We estimate mass-loss rates between 1.0 * $10^9$ g/s and 9.8 * $10^9$ g/s for TOI-5108 b and between 3.6 * $10^8$ g/s and 3.5 * $10^9$ g/s for TOI-5786 b. Given their masses, this means that both planets are stable against photoevaporation. We also detect a transit signal for a second planet candidate in the TESS data of TOI-5786 with a period of 6.998 days and a radius of $3.83 \pm 0.16$ $R_\oplus$. Using our RV data and photodynamical modeling, we are able to provide a 3-$\sigma$ upper limit of 26.5 $M_\oplus$ for the mass of the potential inner companion to TOI-5786 b., Comment: 25 pages, 13 figures. Accepted for publication in A&A

The James Webb Space Telescope (JWST) is unveiling the rest-frame near-IR structure of galaxies. We measure the evolution with redshift of the rest-frame optical and near-IR S\'ersic index ($n$), and examine the dependence on stellar mass and star-formation activity across the redshift range $0.5\leq z\leq2.5$. We infer rest-frame near-IR S\'ersic profiles for $\approx 15.000$ galaxies in publicly available NIRCam imaging mosaics from the COSMOS-Web and PRIMER surveys. We augment these with rest-frame optical S\'ersic indices, previously measured from HST imaging mosaics. The median S\'ersic index evolves slowly or not at all with redshift, except for very high-mass galaxies ($M_\star > 10^{11}~{\text{M}}_\odot$), which show an increase from $n\approx 2.5$ to $n\approx 4$ at $z<1$. High-mass galaxies have higher $n$ than lower-mass galaxies ($M_\star=10^{9.5}~{\text{M}}_\odot$) at all redshifts, with a stronger dependence in the rest-frame near-IR than in the rest-frame optical at $z>1$. This wavelength dependence is caused by star-forming galaxies that have lower optical than near-IR $n$ at z>1 (but not at z<1). Both at optical and near-IR wavelengths, star-forming galaxies have lower $n$ than quiescent galaxies, fortifying the connection between star-formation activity and radial stellar mass distribution. At $z>1$ the median near-IR $n$ varies strongly with star formation activity, but not with stellar mass. The scatter in near-IR $n$ is higher in the green valley (0.25 dex) than on the star-forming sequence and among quiescent galaxies (0.18 dex) -- this trend is not seen in the optical because dust and young stars contribute to the variety in optical light profiles. Our newly measured rest-frame near-IR radial light profiles motivate future comparisons with radial stellar mass profiles of simulated galaxies as a stringent constraint on processes that govern galaxy formation., Comment: Accepted for publication on A&A

Large Language Models (LLMs) have been shown to be susceptible to jailbreak attacks, or adversarial attacks used to illicit high risk behavior from a model. Jailbreaks have been exploited by cybercriminals and blackhat actors to cause significant harm, highlighting the critical need to safeguard widely-deployed models. Safeguarding approaches, which include fine-tuning models or having LLMs "self-reflect", may lengthen the inference time of a model, incur a computational penalty, reduce the semantic fluency of an output, and restrict ``normal'' model behavior. Importantly, these Safety-Performance Trade-offs (SPTs) remain an understudied area. In this work, we introduce a novel safeguard, called SafeNudge, that combines Controlled Text Generation with "nudging", or using text interventions to change the behavior of a model. SafeNudge triggers during text-generation while a jailbreak attack is being executed, and can reduce successful jailbreak attempts by 30% by guiding the LLM towards a safe responses. It adds minimal latency to inference and has a negligible impact on the semantic fluency of outputs. Further, we allow for tunable SPTs. SafeNudge is open-source and available through https://pypi.org/, and is compatible with models loaded with the Hugging Face "transformers" library.

Continuous gravitational waves (CWs) emission from neutron stars carries information about their internal structure and equation of state, and it can provide tests of General Relativity. We present a search for CWs from a set of 45 known pulsars in the first part of the fourth LIGO--Virgo--KAGRA observing run, known as O4a. We conducted a targeted search for each pulsar using three independent analysis methods considering the single-harmonic and the dual-harmonic emission models. We find no evidence of a CW signal in O4a data for both models and set upper limits on the signal amplitude and on the ellipticity, which quantifies the asymmetry in the neutron star mass distribution. For the single-harmonic emission model, 29 targets have the upper limit on the amplitude below the theoretical spin-down limit. The lowest upper limit on the amplitude is $6.4\!\times\!10^{-27}$ for the young energetic pulsar J0537-6910, while the lowest constraint on the ellipticity is $8.8\!\times\!10^{-9}$ for the bright nearby millisecond pulsar J0437-4715. Additionally, for a subset of 16 targets we performed a narrowband search that is more robust regarding the emission model, with no evidence of a signal. We also found no evidence of non-standard polarizations as predicted by the Brans-Dicke theory., Comment: main paper: 12 pages, 6 figures, 4 tables

In high energy density physics (HEDP) and inertial confinement fusion (ICF), predictive modeling is complicated by uncertainty in parameters that characterize various aspects of the modeled system, such as those characterizing material properties, equation of state (EOS), opacities, and initial conditions. Typically, however, these parameters are not directly observable. What is observed instead is a time sequence of radiographic projections using X-rays. In this work, we define a set of sparse hydrodynamic features derived from the outgoing shock profile and outer material edge, which can be obtained from radiographic measurements, to directly infer such parameters. Our machine learning (ML)-based methodology involves a pipeline of two architectures, a radiograph-to-features network (R2FNet) and a features-to-parameters network (F2PNet), that are trained independently and later combined to approximate a posterior distribution for the parameters from radiographs. We show that the estimated parameters can be used in a hydrodynamics code to obtain density fields and hydrodynamic shock and outer edge features that are consistent with the data. Finally, we demonstrate that features resulting from an unknown EOS model can be successfully mapped onto parameters of a chosen analytical EOS model, implying that network predictions are learning physics, with a degree of invariance to the underlying choice of EOS model.

Deep learning methods are highly effective for many image reconstruction tasks. However, the performance of supervised learned models can degrade when applied to distinct experimental settings at test time or in the presence of distribution shifts. In this study, we demonstrate that pruning deep image reconstruction networks at training time can improve their robustness to distribution shifts. In particular, we consider unrolled reconstruction architectures for accelerated magnetic resonance imaging and introduce a method for pruning unrolled networks (PUN) at initialization. Our experiments demonstrate that when compared to traditional dense networks, PUN offers improved generalization across a variety of experimental settings and even slight performance gains on in-distribution data., Comment: 5 pages, 4 figures, Asilomar Conference on Signals, Systems, and Computers 2024

To understand quantum optics experiments, we must perform calculations that consider the principal sources of noise, such as losses, spectral impurity and partial distinguishability. In both discrete and continuous variable systems, these can be modeled as mixed Gaussian states over multiple modes. The modes are not all resolved by photon-number measurements and so require calculations on coarse-grained photon-number distribution. Existing methods can lead to a combinatorial explosion in the time complexity, making this task unfeasible for even moderate sized experiments of interest. In this work, we prove that the computation of this type of distributions can be done in exponential time, providing a combinatorial speedup. We develop numerical techniques that allow us to determine coarse-grained photon number distributions of Gaussian states, as well as density matrix elements of heralded non-Gaussian states prepared in the presence of spectral impurity and partial distinguishability. These results offer significant speed-up and accuracy improvements to validation tests of both Fock and Gaussian boson samplers that rely on binned probability distributions. Moreover, they pave the way to a more efficient simulation of realistic photonic circuits, unlocking the ability to perform exact calculations at scales which were previously out of reach. In addition to this, our results, including loop Hafnian master theorems, may be of interest to the fields of combinatorics and graph theory., Comment: 29 pages, 9 figures

Large language models (LLMs) are often fine-tuned for use on downstream tasks, though this can degrade capabilities learned during previous training. This phenomenon, often referred to as catastrophic forgetting, has important potential implications for the safety of deployed models. In this work, we first show that models trained on downstream tasks forget their safety tuning to a greater extent than models trained in the opposite order. Second, we show that forgetting disproportionately impacts safety information about certain groups. To quantify this phenomenon, we define a new metric we term biased forgetting. We conduct a systematic evaluation of the effects of task ordering on forgetting and apply mitigations that can help the model recover from the forgetting observed. We hope our findings can better inform methods for chaining the finetuning of LLMs in continual learning settings to enable training of safer and less toxic models.

Concerns about the risks and harms posed by artificial intelligence (AI) have resulted in significant study into algorithmic transparency, giving rise to a sub-field known as Explainable AI (XAI). Unfortunately, despite a decade of development in XAI, an existential challenge remains: progress in research has not been fully translated into the actual implementation of algorithmic transparency by organizations. In this work, we test an approach for addressing the challenge by creating transparency advocates, or motivated individuals within organizations who drive a ground-up cultural shift towards improved algorithmic transparency. Over several years, we created an open-source educational workshop on algorithmic transparency and advocacy. We delivered the workshop to professionals across two separate domains to improve their algorithmic transparency literacy and willingness to advocate for change. In the weeks following the workshop, participants applied what they learned, such as speaking up for algorithmic transparency at an organization-wide AI strategy meeting. We also make two broader observations: first, advocacy is not a monolith and can be broken down into different levels. Second, individuals' willingness for advocacy is affected by their professional field. For example, news and media professionals may be more likely to advocate for algorithmic transparency than those working at technology start-ups.

We perform an all-order analysis of double-logarithmic corrections to the so-called soft-overlap contribution to heavy-to-light transition form factors at large hadronic recoil. Specifically, we study $B_c \to \eta_c$ transitions within a perturbative non-relativistic framework, treating both the bottom and charm quarks as heavy with the hierarchy $m_b \gg m_c \gg\Lambda_{\rm QCD}$. Our diagrammatic analysis shows that double-logarithmic corrections arise from two distinct sources: Exponentiated soft-gluon effects described by standard Sudakov factors, and rapidity-ordered soft-quark configurations, leading to implicit integral equations, which so far have only been studied in the context of energetic muon-electron backward scattering. We find that the all-order structure of the double logarithms is governed by a novel type of coupled integral equations, which encode the non-trivial interplay between these two effects. Whereas a closed-form solution to these equations is currently unknown, we present useful iteration formulas, and derive the asymptotic behaviour of the soft-overlap form factor for infinitely large recoil energies, showing that the Sudakov suppression is somewhat weakened by the intertwined soft-quark and soft-gluon corrections. In a broader context, our findings shed light onto the physical origin and mathematical structure of endpoint divergences arising from soft-collinear factorization and the related Feynman mechanism for power-suppressed hard exclusive processes., Comment: 38 pages, 9 figures

The first JWST/MIRI photometric observations of TRAPPIST-1 b allowed for the detection of the thermal emission of the planet at 15 $\mu m$, suggesting that the planet could be a bare rock with a zero albedo and no redistribution of heat. These observations at 15 $\mu m$ were acquired as part of GTO time that included a twin program at 12.8 $\mu m$ in order to have a measurement in and outside the CO$_2$ absorption band. Here we present five new occultations of TRAPPIST-1 b observed with MIRI in an additional photometric band at 12.8 $\mu m$. We perform a global fit of the 10 eclipses and derive a planet-to-star flux ratio and 1-$\sigma$ error of 452 $\pm$ 86 ppm and 775 $\pm$ 90 ppm at 12.8 $\mu m$ and 15 $\mu m$, respectively. We find that two main scenarios emerge. An airless planet model with an unweathered (fresh) ultramafic surface, that could be indicative of relatively recent geological processes fits well the data. Alternatively, a thick, pure-CO2 atmosphere with photochemical hazes that create a temperature inversion and result in the CO2 feature being seen in emission also works, although with some caveats. Our results highlight the challenges in accurately determining a planet's atmospheric or surface nature solely from broadband filter measurements of its emission, but also point towards two very interesting scenarios that will be further investigated with the forthcoming phase curve of TRAPPIST-1 b., Comment: 49 pages, 3 main text figure, 2 extended figures, 10 supplementary figures, accepted for publication in Nature Astronomy on October 29, 2024

We present the study of seven systems, three of which TOI-2295, TOI-2537, and TOI-5110 are newly discovered planetary systems. Through the analysis of TESS photometry, SOPHIE radial velocities, and high-spatial resolution imaging, we found that TOI-2295b, TOI-2537b, and TOI-5110b are transiting warm Jupiters with orbital periods ranging from 30 to 94 d, masses between 0.9 and 2.9 MJ, and radii ranging from 1.0 to 1.5 RJ. Both TOI-2295 and TOI-2537 each harbor at least one additional, outer planet. Their outer planets TOI-2295c and TOI-2537c are characterized by orbital periods of 966.5 +/- 4.3 and 1920^{+230}_{-140} d, respectively, and minimum masses of 5.61^{+0.23}_{-0.24} and 7.2 +/- 0.5 MJ, respectively. We also investigated and characterized the two recently reported warm Jupiters TOI-1836b and TOI-5076b, which we independently detected in SOPHIE RVs. Additionally, we study the planetary candidates TOI-4081.01 and TOI-4168.01. For TOI-4081.01, despite our detection in radial velocities, we cannot rule out perturbation by a blended eclipsing binary and thus exercise caution regarding its planetary nature. On the other hand, we identify TOI-4168.01 as a firm false positive. Finally, we highlight interesting characteristics of these new planetary systems. The transits of TOI-2295b are highly grazing, with an impact parameter of 1.056$^{+0.063}_{-0.043}$. TOI-2537b, in turn, is a temperate Jupiter with an effective temperature of 307+/-15 K and can serve as a valuable low-irradiation control for models of hot Jupiter inflation anomalies. We also detected significant transit timing variations (TTVs) for TOI-2537b, which are likely caused by gravitational interactions with the outer planet TOI-2537c. Finally, TOI-5110b stands out due to its orbital eccentricity of 0.75+/- 0.03, one of the highest planetary eccentricities discovered thus far., Comment: 46 pages, 21 figures, accepted for publication in A&A

It is essential that mesoscopic simulations of reactive systems reproduce the correct statistical distributions at thermodynamic equilibrium. By considering a compressible fluctuating hydrodynamics (FHD) simulation method of ideal gas mixtures undergoing reversible reactions described by the chemical Langevin equations, we show that thermodynamic consistency in reaction rates and the use of instantaneous temperatures for the evaluation of reaction rates is required for fluctuations for the overall system to be correct. We then formulate the required properties of a thermodynamically-consistent reaction (TCR) model. As noted in the literature, while reactions are often discussed in terms of forward and reverse rates, these rates should not be modeled independently because they must be compatible with thermodynamic equilibrium for the system. Using a simple TCR model where each chemical species has constant heat capacity, we derive the explicit condition that the forward and reverse reaction rate constants must satisfy in order for the system to be thermodynamically consistent. We perform equilibrium and non-equilibrium simulations of ideal gas mixtures undergoing a reversible dimerization reaction to measure the fluctuational behavior of the system numerically. We confirm that FHD simulations with the TCR model give the correct static structure factor of equilibrium fluctuations. For the statistically steady simulation of a gas mixture between two isothermal walls with different temperatures, we show using the TCR model that the temperature variance agrees with the corresponding thermodynamic-equilibrium temperature variance in the interior of the system, whereas noticeable deviations are present in regions near walls, where chemistry is far from equilibrium.

Advances in artificial intelligence (AI) present significant risks and opportunities, requiring improved governance to mitigate societal harms and promote equitable benefits. Current incentive structures and regulatory delays may hinder responsible AI development and deployment, particularly in light of the transformative potential of large language models (LLMs). To address these challenges, we propose developing the following three contributions: (1) a large multimodal text and economic-timeseries foundation model that integrates economic and natural language policy data for enhanced forecasting and decision-making, (2) algorithmic mechanisms for eliciting diverse and representative perspectives, enabling the creation of data-driven public policy recommendations, and (3) an AI-driven web platform for supporting transparent, inclusive, and data-driven policymaking.

Dementia is associated with various cognitive impairments and typically manifests only after significant progression, making intervention at this stage often ineffective. To address this issue, the Prediction and Recognition of Cognitive Decline through Spontaneous Speech (PROCESS) Signal Processing Grand Challenge invites participants to focus on early-stage dementia detection. We provide a new spontaneous speech corpus for this challenge. This corpus includes answers from three prompts designed by neurologists to better capture the cognition of speakers. Our baseline models achieved an F1-score of 55.0% on the classification task and an RMSE of 2.98 on the regression task., Comment: 2 pages, no figure, conference

In the last decade, we have been able to probe further down the galaxy luminosity function than ever before and expand into the regime of ultra-faint dwarfs (UFDs), which are some of the best probes we have of small-scale cosmology and galaxy formation. Digital sky surveys have enabled the discovery and study of these incredibly low-mass, highly dark-matter dominated systems around the Local Group, but it is critical that we expand the satellite census further out to understand if Milky Way and M31 satellites are representative of dwarf populations in the local Universe. Using data from HST/ACS, we present updated characterization of four satellite systems in the M81 group. These systems - D1005+68, D1006+69, DWJ0954+6821, and D1009+68 - were previously discovered using ground-based Subaru HSC data as overdensities in M81's halo and are now confirmed with HST/ACS by this work. These are all faint (M_V >= -7.9) and consistent with old (~13 Gyr), metal-poor ([M/H] < -1.5) populations. Each system possesses relatively unusual features - including one of the most concentrated satellite galaxies with a Sersic index of n ~ 5, one of the most elliptical galaxies outside the Local Group with an e ~ 0.6, and one of the most compact galaxies for its magnitude. Two of the satellites have very low surface brightness, lower than most known galaxies in this absolute magnitude range. This work previews the scientific promise of the upcoming Rubin Observatory and Roman Telescope for illuminating the diversity of UFDs in the Local Volume and beyond., Comment: 24 pages, 12 figures, 1 appendix, accepted to ApJ

Large-scale neutrino detectors currently under construction will have the unique ability to probe the annihilation of low-mass thermal-relic dark matter to neutrinos. This represents an essential test of the thermal freezeout paradigm. This raises the question: what viable UV-complete models are there in which dark matter annihilates dominantly to neutrinos? We discuss models that fulfill this criteria, and are invariant under the Standard Model gauge group, for both scalar and fermionic dark matter. Specifically, we construct new models in which annihilation via the $s$-channel exchange of a scalar or pseudoscalar mediator achieves the correct relic density. In these models, dark matter is stabilised by an exact or a softly-broken lepton-number symmetry. The parameter space of such models will be probed, almost entirely, by the combination of JUNO, Hyper-Kamiokande and CMB-S4., Comment: 10 pages, 3 figures, 3 tables

Photon upconversion, combining several low-energy photons to generate one high-energy photon is of wide interest for biomedical, catalytic and photonic applications. Lanthanide-doped nanoparticles (LnNP) are a unique type of upconversion nanoconverter, which can realize ultralarge anti-Stokes shift (>1000 nm) and high photostability, without photo-bleaching and photo-blinking. The excitation wavelength of LnNPs has been limited to the second near-infrared window (1000-1700 nm), mainly sensitized by erbium ions with absorption centered around 1.5 $\mu$m. Here, we demonstrate novel self-sensitized holmium (Ho)-doped nanoconverters to further expand the sensitization range to the short-wave infrared at 2 $\mu$m and achieve efficient upconversion to 640 nm. We show that this upconversion is a 4-photon conversion process with an underlying energy transfer upconversion mechanism. Via careful control of dopant concentration and shelling we achieve a relative upconversion-to-downconversion efficiency up to 15.2%, more than half the theoretical maximum. The placement of the Ho doped LnNPs into a plasmonic nanocavity device enables large gains in emission intensity (up to 32-fold), due to the dramatic shortening of the emission lifetime of Ho from 29 $\mu$s to <1 ns, indicating a high Purcell-enhancement factor of 3x10$^4$. These results open new possibilities at the frontier of short-wave infrared upconversion and the nanoplasmonic enhancement of LnNP emission, with potential applications in detection, theranostics, photonics and optoelectronics.

Let $\Bbbk$ be a field of characteristic zero. Motivated by the fundamental question of whether it is possible for the universal enveloping algebra of an infinite-dimensional Lie algebra to be noetherian, we study Lie algebras of derivations of associative algebras. The main result of this paper is that the universal enveloping algebra of the Lie algebra of derivations of a finitely generated $\Bbbk$-algebra is not noetherian. This extends a result of Sierra and Walton on the Witt algebra, as well as a result of the second author on Krichever-Novikov algebras. We highlight that the result applies to derivations of both commutative and noncommutative algebras without restriction on their growth., Comment: 12 pages. Comments welcome!

The search for radio technosignatures is an anomaly detection problem: candidate signals represent needles of interest in the proverbial haystack of radio-frequency interference (RFI). Current search frameworks find an enormity of false-positive signals, especially in large surveys, requiring manual follow-up to a sometimes prohibitive degree. Unsupervised learning provides an algorithmic way to winnow the most anomalous signals from the chaff, as well as group together RFI signals that bear morphological similarities. We present GLOBULAR (Grouping Low-frequency Observations By Unsupervised Learning After Reduction) clustering, a signal processing method that uses HDBSCAN to reduce the false-positive rate and isolate outlier signals for further analysis. When combined with a standard narrowband signal detection and spatial filtering pipeline, such as turboSETI, GLOBULAR clustering offers significant improvements in the false-positive rate over the standard pipeline alone, suggesting dramatic potential for the amelioration of manual follow-up requirements for future large surveys. By removing RFI signals in regions of high spectral occupancy, GLOBULAR clustering may also enable the detection of signals missed by the standard pipeline. We benchmark our method against the Choza et al. (2024) turboSETI-only search of 97 nearby galaxies at L-band, demonstrating a false-positive hit reduction rate of 93.1% and a false-positive event reduction rate of 99.3%., Comment: 20 pages, 14 figures, submitted to AJ

We provide a formula for the order of the Tate--Shafarevich group of elliptic curves over dihedral extensions of number fields of order $2n$, up to $4^{th}$ powers and primes dividing $n$. Specifically, for odd $n$ it is equal to the order of the Tate--Shafarevich group over the quadratic subextension. A similar formula holds for even $n$.

We present a distributed algorithm to track the fixed points of time-varying quadratic programs when agents can (i) sample their objective function asynchronously, (ii) compute new iterates asynchronously, and (iii) communicate asynchronously. We show that even for a time-varying strongly convex quadratic program, asynchronous sampling of objectives can cause agents to minimize a certain form of nonconvex "aggregate" objective function. Then, we show that by minimizing the aggregate objective, agents still track the solution of the original quadratic program to within an error ball dependent upon (i) agents' tracking error when solving the aggregate problem, and (ii) the time variation of the original quadratic objective. Lastly, we show that this work generalizes existing work, in the sense that synchronizing the agents' behaviors recovers existing convergence results (up to constants). Simulations show the robustness of our algorithm to asynchrony.

The rapid proliferation of data centers in the US - driven partly by the adoption of artificial intelligence - has set off alarm bells about the industry's environmental impact. We compiled detailed information on 2,132 US data centers operating between September 2023 and August 2024 and determined their electricity consumption, electricity sources, and attributable CO$_{2}$e emissions. Our findings reveal that data centers accounted for more than 4% of total US electricity consumption - with 56% derived from fossil fuels - generating more than 105 million tons of CO$_{2}$e (2.18% of US emissions in 2023). Data centers' carbon intensity - the amount of CO$_{2}$e emitted per unit of electricity consumed - exceeded the US average by 48%. Our data pipeline and visualization tools can be used to assess current and future environmental impacts of data centers.

We report the discovery of a super-Earth candidate orbiting the nearby mid M dwarf Gl\,725A using the radial velocity (RV) method. The planetary signal has been independently identified using high-precision RVs from the SOPHIE and SPIRou spectrographs, in the optical and near-infrared domains, respectively. We modelled the stellar activity signal jointly with the planet using two Gaussian Processes, one for each instrument to account for the chromaticity of the stellar activity and instrumental systematics, along with a Keplerian model. The signal is significantly detected with a RV semi-amplitude of $1.67\pm0.20$ m/s. The planet Gl 725A b is found to be in an orbit compatible with circular with a period of $11.2201\pm0.0051$ days. We analysed 27 sectors of TESS photometry on which no transit event was found. We determined a minimum mass of $M_{p}\sin{i}=2.78\pm0.35\,M_{\oplus}$ which places the planet in the super-Earth regime. Using Mass-Radius relationships we predict a planetary radius to be between 1.2 and $2.0\,R_{\oplus}$. The proximity of Gl 725A, of only 3.5 pc, makes this new exoplanet one of the closest to Earth and joins the group of S-type low-mass planets in short orbits ($P<15$ d) around close M dwarfs., Comment: 26 pages, 17 figures. Accepted in A&A

We simulate the mesoscopic dynamics of droplets formed by phase separated fluids at nanometer scales where thermal fluctuations are significant. Both spherical droplets fully immersed in a second fluid and sessile droplets which are also in contact with a solid surface are studied. Our model combines a Cahn-Hillard formulation with incompressible fluctuating hydrodynamics; for sessile droplets the fluid-solid contact angle is specified as a boundary condition. Deterministic simulations with an applied body force are used to measure the droplets' mobility from which a diffusion coefficient is obtained using the Einstein relation. Stochastic simulations are independently used to obtain a diffusion coefficient from a linear fit of the variance of a droplet's position with time. In some scenarios these two measurements give the same value but not in the case of a spherical droplet initialized near a slip wall or in the case of sessile droplets with large contact angles (greater than 90 degrees) on both slip and no-slip surfaces.

Text toxicity detection systems exhibit significant biases, producing disproportionate rates of false positives on samples mentioning demographic groups. But what about toxicity detection in speech? To investigate the extent to which text-based biases are mitigated by speech-based systems, we produce a set of high-quality group annotations for the multilingual MuTox dataset, and then leverage these annotations to systematically compare speech- and text-based toxicity classifiers. Our findings indicate that access to speech data during inference supports reduced bias against group mentions, particularly for ambiguous and disagreement-inducing samples. Our results also suggest that improving classifiers, rather than transcription pipelines, is more helpful for reducing group bias. We publicly release our annotations and provide recommendations for future toxicity dataset construction.

We address the question of how to connect predictions by hydrodynamic models of how sea lice move in water to observable measures that count the number of lice on each fish in a cage in the water. This question is important for management and regulation of aquacultural practice that tries to maximise food production and minimise risk to the environment. We do this through a simple rule-based model of interaction between sea lice and caged fish. The model is simple: sea lice can attach and detach from a fish. The model has a novel feature, encoding what is known as a master equation producing a time-series of distributions of lice on fish that one might expect to find if a cage full of fish were placed at any given location. To demonstrate how this works, and to arrive at a rough estimate of the interaction rates, we fit a simplified version of the model with three free parameters to publicly available data about an experiment with sentinel cages in Loch Linnhe in Scotland. Our construction, coupled to the hydrodynamic models driven by surveillance data from industrial farms, quantifies the environmental impact as: what would the infection burden look like in a notional cage at any location and how does it change with time?

Van der Waals (vdW) materials exhibit a variety of states that can be switched with low power at low temperatures, offering a viable cryogenic "flash memory" required for the classical control electronics for solid-state quantum information processing. In 1T-TaS2, a non-volatile metallic 'hidden' state can be induced from an insulating equilibrium charge-density wave ground state using either optical or electrical pulses. Given that conventional memristors form localized, filamentary channels which support the current, a key question for design concerns the geometry of the conduction region in highly energy-efficient 1T-TaS2 devices. Here, we report in operando micro-beam X-ray diffraction, fluorescence, and concurrent transport measurements, allowing us to spatially image the non-thermal hidden state induced by electrical switching of 1T-TaS2. Our results reveal a long-range ordered, non-filamentary switched state that extends well below the electrodes, implying that the self-organized, collective growth of the hidden phase is driven by a combination of charge flow and lattice strain. Our unique combination of techniques showcases the potential of non-destructive, three-dimensional X-ray imaging to study bulk switching properties in microscopic detail, namely electrical control of the vdW layer stacking.

Learning correlations from data forms the foundation of today's machine learning (ML) and artificial intelligence (AI) research. While such an approach enables the automatic discovery of patterned relationships within big data corpora, it is susceptible to failure modes when unintended correlations are captured. This vulnerability has expanded interest in interrogating spuriousness, often critiqued as an impediment to model performance, fairness, and robustness. In this article, we trace deviations from the conventional definition of statistical spuriousness-which denotes a non-causal observation arising from either coincidence or confounding variables-to articulate how ML researchers make sense of spuriousness in practice. Drawing on a broad survey of ML literature, we conceptualize the "multiple dimensions of spuriousness," encompassing: relevance ("Models should only use correlations that are relevant to the task."), generalizability ("Models should only use correlations that generalize to unseen data"), human-likeness ("Models should only use correlations that a human would use to perform the same task"), and harmfulness ("Models should only use correlations that are not harmful"). These dimensions demonstrate that ML spuriousness goes beyond the causal/non-causal dichotomy and that the disparate interpretative paths researchers choose could meaningfully influence the trajectory of ML development. By underscoring how a fundamental problem in ML is contingently negotiated in research contexts, we contribute to ongoing debates about responsible practices in AI development.

Robot learning holds tremendous promise to unlock the full potential of flexible, general, and dexterous robot systems, as well as to address some of the deepest questions in artificial intelligence. However, bringing robot learning to the level of generality required for effective real-world systems faces major obstacles in terms of data, generalization, and robustness. In this paper, we discuss how generalist robot policies (i.e., robot foundation models) can address these challenges, and how we can design effective generalist robot policies for complex and highly dexterous tasks. We propose a novel flow matching architecture built on top of a pre-trained vision-language model (VLM) to inherit Internet-scale semantic knowledge. We then discuss how this model can be trained on a large and diverse dataset from multiple dexterous robot platforms, including single-arm robots, dual-arm robots, and mobile manipulators. We evaluate our model in terms of its ability to perform tasks in zero shot after pre-training, follow language instructions from people and from a high-level VLM policy, and its ability to acquire new skills via fine-tuning. Our results cover a wide variety of tasks, such as laundry folding, table cleaning, and assembling boxes., Comment: See project website for videos: https://physicalintelligence.company/blog/pi0

We report on the fabrication and characterization of superconducting magnesium diboride (MgB$_2$) thin films intended for quantum-limited devices based non-linear kinetic inductance (NLKI) such as parametric amplifiers with either elevated operating temperatures or expanded frequency ranges. In order to characterize the MgB$_2$ material properties, we have fabricated coplanar waveguide (CPW) transmission lines and microwave resonators using $\approx$ 40 nm thick MgB$_2$ films with a measured kinetic inductance of $\sim$ 5.5 pH/$\square$ and internal quality factors, $Q_i \approx 3 \times 10^4$ at 4.2 K. We measure the NLKI in MgB$_2$ by applying a DC bias to a 6 cm long by 4 $\mu$m wide CPW transmission line, and measuring the resulting phase delay caused by the current dependent NLKI. We also measure the current dependent NLKI through CPW resonators that shift down in frequency with increased power applied through the CPW feedline. Using these measurements, we calculate the characteristic non-linear current parameter, $I_*$, for multiple CPW geometries. We find values for corresponding current density, $J_* = 12-22$~MA/cm$^2$ and a ratio of the critical current to the non-linear current parameter, $I_C/I_* = 0.14-0.26$, similar to or higher than values for other superconductors such as NbTiN and TiN.

The Local Ultraviolet to Infrared Treasury (LUVIT) is a Hubble Space Telescope program that combines newly acquired data in the near ultraviolet (NUV), optical, and near infrared (NIR) with archival optical and NIR imaging to produce multiband panchromatic resolved stellar catalogs for 23 pointings in 22 low-mass, star-forming galaxies ranging in distance from the outskirts of the Local Group to ~3.8 Mpc. We describe the survey design, detail the LUVIT broadband filter observations and the archival datasets included in the LUVIT reductions, and summarize the simultaneous multiband data reduction steps. The spatial distributions and color-magnitude diagrams (CMDs) from the resulting stellar catalogs are presented for each target, from the NUV to the NIR. We demonstrate in which regions of the CMDs stars with NUV and optical, optical and NIR, and NUV through NIR detections reside. For each target, we use the results from artificial star tests to measure representative completeness, bias, and total photometric uncertainty as a function of magnitude in each broadband filter. We also assess which LUVIT targets have significant spatial variation in the fraction of stars recovered at a given magnitude. The panchromatic LUVIT stellar catalogs will provide a rich legacy dataset for a host of resolved stellar population studies., Comment: 48 pages, 14 figures, 8 tables, published in ApJS

The XLZD collaboration is developing a two-phase xenon time projection chamber with an active mass of 60 to 80 t capable of probing the remaining WIMP-nucleon interaction parameter space down to the so-called neutrino fog. In this work we show that, based on the performance of currently operating detectors using the same technology and a realistic reduction of radioactivity in detector materials, such an experiment will also be able to competitively search for neutrinoless double beta decay in $^{136}$Xe using a natural-abundance xenon target. XLZD can reach a 3$\sigma$ discovery potential half-life of 5.7$\times$10$^{27}$ yr (and a 90% CL exclusion of 1.3$\times$10$^{28}$ yr) with 10 years of data taking, corresponding to a Majorana mass range of 7.3-31.3 meV (4.8-20.5 meV). XLZD will thus exclude the inverted neutrino mass ordering parameter space and will start to probe the normal ordering region for most of the nuclear matrix elements commonly considered by the community., Comment: 29 pages, 7 figures

This report describes the experimental strategy and technologies for a next-generation xenon observatory sensitive to dark matter and neutrino physics. The detector will have an active liquid xenon target mass of 60-80 tonnes and is proposed by the XENON-LUX-ZEPLIN-DARWIN (XLZD) collaboration. The design is based on the mature liquid xenon time projection chamber technology of the current-generation experiments, LZ and XENONnT. A baseline design and opportunities for further optimization of the individual detector components are discussed. The experiment envisaged here has the capability to explore parameter space for Weakly Interacting Massive Particle (WIMP) dark matter down to the neutrino fog, with a 3$\sigma$ evidence potential for the spin-independent WIMP-nucleon cross sections as low as $3\times10^{-49}\rm cm^2$ (at 40 GeV/c$^2$ WIMP mass). The observatory is also projected to have a 3$\sigma$ observation potential of neutrinoless double-beta decay of $^{136}$Xe at a half-life of up to $5.7\times 10^{27}$ years. Additionally, it is sensitive to astrophysical neutrinos from the atmosphere, sun, and galactic supernovae., Comment: 32 pages, 14 figures