Search

Ranking a set of items based on their relevance to a given query is a core problem in search and recommendation. Transformer-based ranking models are the state-of-the-art approaches for such tasks, but they score each query-item independently, ignoring the joint context of other relevant items. This leads to sub-optimal ranking accuracy and high computational costs. In response, we propose Cross-encoders with Joint Efficient Modeling (CROSS-JEM), a novel ranking approach that enables transformer-based models to jointly score multiple items for a query, maximizing parameter utilization. CROSS-JEM leverages (a) redundancies and token overlaps to jointly score multiple items, that are typically short-text phrases arising in search and recommendations, and (b) a novel training objective that models ranking probabilities. CROSS-JEM achieves state-of-the-art accuracy and over 4x lower ranking latency over standard cross-encoders. Our contributions are threefold: (i) we highlight the gap between the ranking application's need for scoring thousands of items per query and the limited capabilities of current cross-encoders; (ii) we introduce CROSS-JEM for joint efficient scoring of multiple items per query; and (iii) we demonstrate state-of-the-art accuracy on standard public datasets and a proprietary dataset. CROSS-JEM opens up new directions for designing tailored early-attention-based ranking models that incorporate strict production constraints such as item multiplicity and latency.

Classification tasks present challenges due to class imbalances and evolving data distributions. Addressing these issues requires a robust method to handle imbalances while effectively detecting out-of-distribution (OOD) samples not encountered during training. This study introduces a novel OOD detection algorithm designed for tabular datasets, titled Deep Neural Network-based Gaussian Descriptor for Imbalanced Tabular Data (DNN-GDITD). The DNN-GDITD algorithm can be placed on top of any DNN to facilitate better classification of imbalanced data and OOD detection using spherical decision boundaries. Using a combination of Push, Score-based, and focal losses, DNN-GDITD assigns confidence scores to test data points, categorizing them as known classes or as an OOD sample. Extensive experimentation on tabular datasets demonstrates the effectiveness of DNN-GDITD compared to three OOD algorithms. Evaluation encompasses imbalanced and balanced scenarios on diverse tabular datasets, including a synthetic financial dispute dataset and publicly available tabular datasets like Gas Sensor, Drive Diagnosis, and MNIST, showcasing DNN-GDITD's versatility., Comment: 17 pages

Ternary Deep Neural Networks (DNN) have shown a large potential for highly energy-constrained systems by virtue of their low power operation (due to ultra-low precision) with only a mild degradation in accuracy. To enable an energy-efficient hardware substrate for such systems, we propose a compute-enabled memory design, referred to as SiTe-CiM, which features computing-in-memory (CiM) of dot products between signed ternary (SiTe) inputs and weights. SiTe CiM is based on cross-coupling of two bit cells to enable CiM of dot products in the signed ternary regime. We explore SiTe CiM with 8T-SRAM, 3T-embedded DRAM (3T-eDRAM) and 3T-ferroelectric metal FET (FEMFET) memories. We propose two flavors of this technique, namely SiTe CiM I/II. In SiTe CiM I, we employ two additional transistors per cell for cross-coupling, achieving fast CiM operations, albeit incurring an area overhead ranging from 18% to 34% (compared to standard ternary memories). In SiTe CiM II, four extra transistors are utilized for every 16 cells in a column, thereby incurring only 6% area cost (but leading to slower CiM than SiTe CiM I). Based on the array analysis, our designs achieve up to 88% lower CiM latency and 78% CiM energy savings across various technologies considered, as compared to their respective near-memory computing counterparts. Further, we perform system level analysis by incorporating SiTe CiM I/II arrays in a ternary DNN accelerator and show up to 7X throughput boost and up to 2.5X energy reduction compared to the near-memory ternary DNN accelerators.

We generalise Phinney's 'practical theorem' to account for modified graviton dispersion relations motivated by certain cosmological scenarios. Focusing on specific examples, we show how such modifications can induce characteristic localised distortions, bumps, in the frequency profile of the stochastic gravitational wave background emitted from distant binary sources. We concentrate on gravitational waves at nano-Hertz frequencies probed by pulsar timing arrays, and we forecast the capabilities of future experiments to accurately probe parameters controlling modified dispersion relations. Our predictions are based on properties of gravitational waves emitted in the first inspiral phase of the binary process, and do not rely on assumptions of non-linear effects occurring during the binary merging phase, Comment: 19 pages, 5 figures, 2 tables

This paper introduces a Scalable Hierarchical Aware Convolutional Neural Network (SHA-CNN) model architecture for Edge AI applications. The proposed hierarchical CNN model is meticulously crafted to strike a balance between computational efficiency and accuracy, addressing the challenges posed by resource-constrained edge devices. SHA-CNN demonstrates its efficacy by achieving accuracy comparable to state-of-the-art hierarchical models while outperforming baseline models in accuracy metrics. The key innovation lies in the model's hierarchical awareness, enabling it to discern and prioritize relevant features at multiple levels of abstraction. The proposed architecture classifies data in a hierarchical manner, facilitating a nuanced understanding of complex features within the datasets. Moreover, SHA-CNN exhibits a remarkable capacity for scalability, allowing for the seamless incorporation of new classes. This flexibility is particularly advantageous in dynamic environments where the model needs to adapt to evolving datasets and accommodate additional classes without the need for extensive retraining. Testing has been conducted on the PYNQ Z2 FPGA board to validate the proposed model. The results achieved an accuracy of 99.34%, 83.35%, and 63.66% for MNIST, CIFAR-10, and CIFAR-100 datasets, respectively. For CIFAR-100, our proposed architecture performs hierarchical classification with 10% reduced computation while compromising only 0.7% accuracy with the state-of-the-art. The adaptability of SHA-CNN to FPGA architecture underscores its potential for deployment in edge devices, where computational resources are limited. The SHA-CNN framework thus emerges as a promising advancement in the intersection of hierarchical CNNs, scalability, and FPGA-based Edge AI.

We investigate the `intrinsic' H$\alpha$ EW distributions of $z \sim 0.4 - 2.2$ narrowband-selected H$\alpha$ samples from HiZELS and DAWN using a forward modeling approach. We find an EW - stellar mass anti-correlation with steepening slopes $-0.18\pm0.03$ to $-0.24^{+0.06}_{-0.08}$ at $z \sim 0.4$ and $z\sim 2.2$, respectively. Typical EW increases as $(1+z)^{1.78^{+0.22}_{-0.23}}$ for a $10^{10}$ M$_\odot$ emitter from $15^{+2.4}_{-2.3}$\r{A} ($z \sim 0.4$) to $67.7^{+10.4}_{-10.0}$\r{A} ($z \sim 2.2$) and is steeper with decreasing stellar mass highlighting the high EW nature of low-mass high-$z$ systems. We model this redshift evolving anti-correlation, $W_0(M,z)$, and find it produces H$\alpha$ luminosity and SFR functions strongly consistent with observations validating the model and allowing us to use $W_0(M,z)$ to investigate the relative contribution of H$\alpha$ emitters towards cosmic SF. We find EW$_0 > 200$ \r{A} emitters contribute significantly to cosmic SF activity at $z \sim 1.5 - 2$ making up $\sim 40$% of total SF consistent with sSFR $> 10^{-8.5}$ yr$^{-1}$ ($\sim 45 - 55$%). Overall, this highlights the importance of high EW systems at high-$z$. Our $W_0(M,z)$ model also reproduces the cosmic sSFR evolution found in simulations and observations and show that tension between the two can simply arise from selection effects in observations. Lastly, we forecast Roman and Euclid grism surveys using $W_0(M,z)$ including observational efficiency and limiting resolution effects where we predict $\sim 24000$ and $\sim 30000$ $0.5 < z < 1.9$ H$\alpha$ emitters per deg$^{-2}$, respectively, down to $>5\times10^{-17}$ erg s$^{-1}$ cm$^{-2}$ including $10^{7.2 - 8}$ M$_\odot$ galaxies at $z > 1$ with EW$_0 >1000$\r{A}. Both Roman and Euclid will enable us to observe with unprecedented detail some of the most bursty/high EW, low-mass star-forming galaxies near cosmic noon., Comment: 25 pages, 12 Figures, and 12 Tables. Submitted to MNRAS. Comments are welcomed. (Abridged Abstract)

Strong emission from doubly ionized oxygen is a beacon for some of the most intensely star forming galaxies known. JWST enables the search for this beacon in the early universe with unprecedented sensitivity. In this work, we extend the study of faint [OIII]$_{5008}$ selected galaxies by an order of magnitude in line luminosity. We use publicly available UNCOVER DR1 JWST/NIRCam and HST imaging data of the cluster lensing field, Abell 2744, to identify strong (rest-frame EW$>500$\AA) [OIII]$_{5008}$ emitters at $z\sim7$ based on excess F410M flux. We find $N=68$ $z\sim7$ [OIII] candidates, with a subset of $N=33$ that have deep HST coverage required to rule-out lower redshift interlopers (13.68 arcmin$^2$ with F814W $5\sigma$ depth $>28$ AB). Such strong emission lines can lead to very red colors that could be misinterpreted as evidence for old, massive stellar populations, but are shown to be due to emission lines where we have spectra. Using this deep HST sample and completeness simulations, which calculate the effective survey volume of the UNCOVER lensing field as a function of [OIII] luminosity, we derive a new [OIII] luminosity function (LF) extending to $41.09<\rm{log}_{10}(L/\rm{erg\,s}^{-1})<42.35$ which is an order of magnitude deeper than previous $z\sim6$ [OIII] LFs based on JWST slitless spectroscopy. This LF is well fit by a power law with a faint-end slope of $\alpha=-2.07^{+0.22}_{-0.23}$. There is little or no evolution between this LF and published [OIII] LFs at redshifts $3\lesssim z\lesssim7$, and no evidence of a turnover at faint luminosities. The sizes of these extreme [OIII] emitters are broadly similar to their low redshift counterparts, the green peas. The luminosity function of [OIII] emitters matches that of Lyman-$\alpha$ at the bright end, suggesting that many of them should be Lyman-$\alpha$ emitters., Comment: submitted to ApJ

Air pollution remains one of the most formidable environmental threats to human health globally, particularly in urban areas, contributing to nearly 7 million premature deaths annually. Megacities, defined as cities with populations exceeding 10 million, are frequent hotspots of severe pollution, experiencing numerous weeks of dangerously poor air quality due to the concentration of harmful pollutants. In addition, the complex interplay of factors makes accurate air quality predictions incredibly challenging, and prediction models often struggle to capture these intricate dynamics. To address these challenges, this paper proposes an attention-enhanced deep multitask spatiotemporal machine learning model based on long-short-term memory networks for long-term air quality monitoring and prediction. The model demonstrates robust performance in predicting the levels of major pollutants such as sulfur dioxide and carbon monoxide, effectively capturing complex trends and fluctuations. The proposed model provides actionable information for policymakers, enabling informed decision making to improve urban air quality., Comment: 6 pages, 3 figures, 3 tables

fmeval is an open source library to evaluate large language models (LLMs) in a range of tasks. It helps practitioners evaluate their model for task performance and along multiple responsible AI dimensions. This paper presents the library and exposes its underlying design principles: simplicity, coverage, extensibility and performance. We then present how these were implemented in the scientific and engineering choices taken when developing fmeval. A case study demonstrates a typical use case for the library: picking a suitable model for a question answering task. We close by discussing limitations and further work in the development of the library. fmeval can be found at https://github.com/aws/fmeval.

In this paper, we explore the feasibility of using machine learning regression as a method of extracting basic stellar parameters and line-of-sight extinctions from spectro-photometric data. We built a stable gradient-boosted random-forest regressor (xgboost), trained on spectroscopic data, capable of producing output parameters with reliable uncertainties from Gaia DR3 data (most notably the low-resolution XP spectra), without ground-based spectroscopic observations. Using Shapley additive explanations, we interpret how the predictions for each star are influenced by each data feature. For the training and testing of the network, we used high-quality parameters obtained from the StarHorse code for a sample of around eight million stars observed by major spectroscopic stellar surveys, complemented by curated samples of hot stars, very metal-poor stars, white dwarfs, and hot sub-dwarfs. The training data cover the whole sky, all Galactic components, and almost the full magnitude range of the Gaia DR3 XP sample of more than 217 million objects that also have reported parallaxes. We have achieved median uncertainties of 0.20 mag in V-band extinction, 0.01 dex in logarithmic effective temperature, 0.20 dex in surface gravity, 0.18 dex in metallicity, and $12\%$ in mass (over the full Gaia DR3 XP sample, with considerable variations in precision as a function of magnitude and stellar type). We succeeded in predicting competitive results based on Gaia DR3 XP spectra compared to classical isochrone or spectral-energy distribution fitting methods we employed in earlier works, especially for parameters $A_V$ and $T_{\rm eff}$, along with the metallicity values. Finally, we showcase some potential applications of this new catalogue, including extinction maps, metallicity trends in the Milky Way, and extended maps of young massive stars, metal-poor stars, and metal-rich stars). [abridged], Comment: A&A, accepted. 13 pages, 13 figures + references & appendices. Data available at https://data.aip.de/projects/shboost2024.html

Background: Incidence of adverse outcome events rises as patients with advanced illness approach end-of-life. Exposures that tend to occur near end-of-life, e.g., use of wheelchair, oxygen therapy and palliative care, may therefore be found associated with the incidence of the adverse outcomes. We propose a strategy for time-to-event analysis to mitigate the time-varying confounding. Methods: We propose a concept of reverse time-to-death (rTTD) and its use for the time-scale in time-to-event analysis. We used data on community-based palliative care uptake (exposure) and emergency department visits (outcome) among patients with advanced cancer in Singapore to illustrate. We compare the results against that of the common practice of using time-on-study (TOS) as time-scale. Results: Graphical analysis demonstrated that cancer patients receiving palliative care had higher rate of emergency department visits than non-recipients mainly because they were closer to end-of-life, and that rTTD analysis made comparison between patients at the same time-to-death. Analysis of emergency department visits in relation to palliative care using TOS time-scale showed significant increase in hazard ratio estimate when observed time-varying covariates were omitted from statistical adjustment (change-in-estimate=0.38; 95% CI 0.15 to 0.60). There was no such change in otherwise the same analysis using rTTD (change-in-estimate=0.04; 95% CI -0.02 to 0.11), demonstrating the ability of rTTD time-scale to mitigate confounding that intensifies in relation to time-to-death. Conclusion: Use of rTTD as time-scale in time-to-event analysis provides a simple and robust approach to control time-varying confounding in studies of advanced illness, even if the confounders are unmeasured., Comment: 22 pages (including 2 tables and 2 figures)

This paper presents a study of the global medium-scale (scales<620 km) gravity wave (GW) activity (in terms of zonal wind variance) and its seasonal, local time and longitudinal variations by employing the enhanced-resolution (~50 km) Whole Atmosphere Model (WAMT254) and space-based observations for geomagnetically quiet conditions. It is found that the GW hotspots produced by WAMT254 in the troposphere and stratosphere agree well with previously well-studied orographic and non-orographic sources. In the ionosphere-thermosphere (IT) region, GWs spread out forming latitudinal band-like hotspots. During solstices, a primary maximum in GW activity is observed in WAMT254 and GOCE over winter mid-high latitudes, likely associated with higher-order waves with primary sources in polar night jet, fronts and polar vortex. During all the seasons, the enhancement of GWs around the geomagnetic poles as observed by GOCE (at ~250 km) is well captured by simulations. WAMT254 GWs in the IT region also show dependence on local time due to their interaction with migrating tides leading to diurnal and semidiurnal variations. The GWs are more likely to propagate up from the MLT region during westward/weakly-eastward phase of thermospheric tides, signifying the dominance of eastward GW momentum flux in the MLT. Additionally, as a novel finding, a wavenumber-4 signature in GW activity is predicted by WAMT254 between 6-12 LT in the tropics at ~250 km, which propagates eastward with local time. This behavior is likely associated with the modulation of GWs by wave-4 signal of non-migrating tides in the lower thermospheric zonal winds.

The Mpemba effect describes the phenomenon that a system at a hot initial temperature cools faster than at an initial warm temperature in the same environment. Such an anomalous cooling has recently been predicted and realized for trapped colloids. Here, we investigate the freezing behavior of a passive colloidal particle by employing numerical Brownian dynamics simulations and theoretical calculations with a model that can be directly tested in experiments. During the cooling process, the colloidal particle exhibits multiple non-monotonic regimes in cooling rates, with the cooling time decreasing twice as a function of the initial temperature-an unexpected phenomenon we refer to as the Double Mpemba effect. Additionally, we demonstrate that both the Mpemba and Double Mpemba effects can be predicted by various machine learning methods, which expedite the analysis of complex, computationally intensive systems.

We searched the Chandra and XMM archives for observations of 900 green pea galaxies to find AGN signatures. Green peas are low-mass galaxies with prominent emission lines, similar in size and star formation rate to high-redshift dwarf galaxies. Of the 29 observations found, 9 show X-ray detections with $S/N>3$. The 2-10 keV X-ray luminosity for these 9 sources exceeds $10^{40}~\mathrm{erg~s}^{-1}$, with 2 sources exceeding $10^{41}~\mathrm{erg~s}^{-1}$, suggesting the presence of intermediate-mass black holes (IMBH) or low-luminosity AGN (LLAGN) with BH masses between $100-10^6M_\mathrm{\odot}$. All X-ray detected sources (plus 6 additional sources) show He~II$\lambda4686$ emission and a broad component of the H$\alpha$ emission line, indicating winds. The line widths of the broad H$\alpha$ and He II$\lambda4686$ emitting gas clouds are weakly correlated ($R^{2}=0.15$), suggesting He II$\lambda4686$ emission is inconsistent with winds from super-Eddington accretors. However, the ratio of X-ray luminosity to star formation rate shows an anti-correlation with metallicity in 5 out of 9 X-ray detected sources, implying ultraluminous X-ray sources are key contributors to the observed X-ray luminosity. This could be due to super-Eddington accretors or IMBH. The X-ray emission is much higher than that produced by Wolf-Rayet stars and supernovae-driven winds. Thus, the X-ray luminosity in these 9 sources can only be explained by black holes with masses over $100~M_\mathrm{\odot}$. Our findings suggest the presence of LLAGN in these galaxies, with broad H$\alpha$ line widths implying BH masses of $10^4-10^6M_\mathrm{\odot}$. Given Green Peas' role as significant Lyman Continuum leakers, LLAGN in these galaxies could have contributed significantly to cosmic reionization., Comment: Submitted to ApJ. 17 pages, 11 figures and 3 tables. Comments welcome

Improving the hardware efficiency of deep neural network (DNN) accelerators with techniques such as quantization and sparsity enhancement have shown an immense promise. However, their inference accuracy in non-ideal real-world settings (such as in the presence of hardware faults) is yet to be systematically analyzed. In this work, we investigate the impact of memory faults on activation-sparse quantized DNNs (AS QDNNs). We show that a high level of activation sparsity comes at the cost of larger vulnerability to faults, with AS QDNNs exhibiting up to 11.13% lower accuracy than the standard QDNNs. We establish that the degraded accuracy correlates with a sharper minima in the loss landscape for AS QDNNs, which makes them more sensitive to perturbations in the weight values due to faults. Based on this observation, we employ sharpness-aware quantization (SAQ) training to mitigate the impact of memory faults. The AS and standard QDNNs trained with SAQ have up to 19.50% and 15.82% higher inference accuracy, respectively compared to their conventionally trained equivalents. Moreover, we show that SAQ-trained AS QDNNs show higher accuracy in faulty settings than standard QDNNs trained conventionally. Thus, sharpness-aware training can be instrumental in achieving sparsity-related latency benefits without compromising on fault tolerance., Comment: arXiv admin note: substantial text overlap with arXiv:2301.00675

We present a logic based interpretable model for learning on graphs and an algorithm to distill this model from a Graph Neural Network (GNN). Recent results have shown connections between the expressivity of GNNs and the two-variable fragment of first-order logic with counting quantifiers (C2). We introduce a decision-tree based model which leverages an extension of C2 to distill interpretable logical classifiers from GNNs. We test our approach on multiple GNN architectures. The distilled models are interpretable, succinct, and attain similar accuracy to the underlying GNN. Furthermore, when the ground truth is expressible in C2, our approach outperforms the GNN., Comment: To Appear in the Proceedings of KR 2024

In health, most large language model (LLM) research has focused on clinical tasks. However, mobile and wearable devices, which are rarely integrated into such tasks, provide rich, longitudinal data for personal health monitoring. Here we present Personal Health Large Language Model (PH-LLM), fine-tuned from Gemini for understanding and reasoning over numerical time-series personal health data. We created and curated three datasets that test 1) production of personalized insights and recommendations from sleep patterns, physical activity, and physiological responses, 2) expert domain knowledge, and 3) prediction of self-reported sleep outcomes. For the first task we designed 857 case studies in collaboration with domain experts to assess real-world scenarios in sleep and fitness. Through comprehensive evaluation of domain-specific rubrics, we observed that Gemini Ultra 1.0 and PH-LLM are not statistically different from expert performance in fitness and, while experts remain superior for sleep, fine-tuning PH-LLM provided significant improvements in using relevant domain knowledge and personalizing information for sleep insights. We evaluated PH-LLM domain knowledge using multiple choice sleep medicine and fitness examinations. PH-LLM achieved 79% on sleep and 88% on fitness, exceeding average scores from a sample of human experts. Finally, we trained PH-LLM to predict self-reported sleep quality outcomes from textual and multimodal encoding representations of wearable data, and demonstrate that multimodal encoding is required to match performance of specialized discriminative models. Although further development and evaluation are necessary in the safety-critical personal health domain, these results demonstrate both the broad knowledge and capabilities of Gemini models and the benefit of contextualizing physiological data for personal health applications as done with PH-LLM., Comment: 72 pages

Despite the proliferation of wearable health trackers and the importance of sleep and exercise to health, deriving actionable personalized insights from wearable data remains a challenge because doing so requires non-trivial open-ended analysis of these data. The recent rise of large language model (LLM) agents, which can use tools to reason about and interact with the world, presents a promising opportunity to enable such personalized analysis at scale. Yet, the application of LLM agents in analyzing personal health is still largely untapped. In this paper, we introduce the Personal Health Insights Agent (PHIA), an agent system that leverages state-of-the-art code generation and information retrieval tools to analyze and interpret behavioral health data from wearables. We curate two benchmark question-answering datasets of over 4000 health insights questions. Based on 650 hours of human and expert evaluation we find that PHIA can accurately address over 84% of factual numerical questions and more than 83% of crowd-sourced open-ended questions. This work has implications for advancing behavioral health across the population, potentially enabling individuals to interpret their own wearable data, and paving the way for a new era of accessible, personalized wellness regimens that are informed by data-driven insights., Comment: 38 pages

The circular polarization of the stochastic gravitational wave background (SGWB) is a key observable for characterising the origin of the signal detected by Pulsar Timing Array (PTA) collaborations. Both the astrophysical and the cosmological SGWB can have a sizeable amount of circular polarization, due to Poisson fluctuations in the source properties for the former, and to parity violating processes in the early universe for the latter. Its measurement is challenging since PTA are blind to the circular polarization monopole, forcing us to turn to anisotropies for detection. We study the sensitivity of current and future PTA datasets to circular polarization anisotropies, focusing on realistic modelling of intrinsic and kinematic anisotropies for astrophysical and cosmological scenarios respectively. Our results indicate that the expected level of circular polarization for the astrophysical SGWB should be within the reach of near future datasets, while for cosmological SGWB circular polarization is a viable target for more advanced SKA-type experiments., Comment: 28 pages, 9 figures

Among the nearly 30,000 known near-Earth asteroids (NEAs), only tens of them possess Earth co-orbital characteristics with semi-major axes $\sim$1 au. In particular, 469219 Kamo`oalewa (2016 HO3), upcoming target of China's Tianwen-2 asteroid sampling mission, exhibits a meta-stable 1:1 mean-motion resonance with Earth. Intriguingly, recent ground-based observations show that Kamo`oalewa has spectroscopic characteristics similar to space-weathered lunar silicates, hinting at a lunar origin instead of an asteroidal one like the vast majority of NEAs. Here we use numerical simulations to demonstrate that Kamo`oalewa's physical and orbital properties are compatible with a fragment from a crater larger than 10--20 km formed on the Moon in the last few million years. The impact could have ejected sufficiently large fragments into heliocentric orbits, some of which could be transferred to Earth 1:1 resonance and persist today. This leads us to suggest the young lunar crater Giordano Bruno (22 km diameter, 1--10 Ma age) as the most likely source, linking a specific asteroid in space to its source crater on the Moon. The hypothesis will be tested by the Tianwen-2 mission when it returns a sample of Kamo`oalewa. And the upcoming NEO Surveyor mission will possibly help us to identify such a lunar-derived NEA population., Comment: 29 pages, 4 figures. Published in Nature Astronomy, 19 April 2024

Dynamical dark energy has gained renewed interest due to recent theoretical and observational developments. In the present paper, we focus on a string-motivated dark energy set-up, and perform a detailed cosmological analysis of exponential quintessence with potential $V=V_0 e^{-\lambda\phi}$, allowing for non-zero spatial curvature. We first gain some physical intuition into the full evolution of such a scenario by analysing the corresponding dynamical system. Then, we test the model using a combination of Planck CMB data, DESI BAO data, as well as recent supernovae datasets. For the model parameter $\lambda$, we obtain a preference for nonzero values: $\lambda = 0.48^{+0.28}_{-0.21},\; 0.68^{+0.31}_{-0.20},\; 0.77^{+0.18}_{-0.15}$ at 68% C.L. when combining CMB+DESI with Pantheon+, Union3 and DES-Y5 supernovae datasets respectively. We find no significant hint for spatial curvature. We discuss the implications of current cosmological results for the exponential quintessence model, and more generally for dark energy in string theory., Comment: 34 pages, 13 figures, 10 tables. Comments welcome! V2 - updated plots and added references

We initiate the study of counting Markov Equivalence Classes (MEC) under logical constraints. MECs are equivalence classes of Directed Acyclic Graphs (DAGs) that encode the same conditional independence structure among the random variables of a DAG model. Observational data can only allow to infer a DAG model up to Markov Equivalence. However, Markov equivalent DAGs can represent different causal structures, potentially super-exponentially many. Hence, understanding MECs combinatorially is critical to understanding the complexity of causal inference. In this paper, we focus on analysing MECs of size one, with logical constraints on the graph topology. We provide a polynomial-time algorithm (w.r.t. the number of nodes) for enumerating essential DAGs (the only members of an MEC of size one) with arbitrary logical constraints expressed in first-order logic with two variables and counting quantifiers (C^2). Our work brings together recent developments in tractable first-order model counting and combinatorics of MECs., Comment: Under Review

The Kepler mission reveals a peculiar trough-peak feature in the orbital spacing of close-in planets near mean-motion resonances: a deficit and an excess that are a couple percent to the narrow, respectively wide, of the resonances. This feature has received two main classes of explanations, one involving eccentricity damping, the other scattering with small bodies. Here, we point out a few issues with the damping scenario, and study the scattering scenario in more detail. We elucidate why scattering small bodies tends to repel two planets. As the small bodies random-walk in energy and angular momentum space, they tend to absorb, fractionally, more energy than angular momentum. This, which we call "ping-pong repulsion", transports angular momentum from the inner to the outer planet and pushes the two planets apart. Such a process, even if ubiquitous, leaves identifiable marks only near first-order resonances: diverging pairs jump across the resonance quickly and produce the MMR asymmetry. To explain the observed positions of the trough-peaks, a total scattering mass of order a few percent of the planet masses is required. Moreover, if this mass is dominated by a handful of Mercury-sized bodies, one can also explain the planet eccentricities as inferred from transit-time-variations. Lastly, we suggest how these conditions may have naturally arisen during the late stages of planet formation., Comment: 14 pages, submitted to AAS journal

We study thermal and non-thermal resonant leptogenesis in a general setting where a heavy scalar $\phi$ decays to right-handed neutrinos (RHNs) whose further out-of-equilibrium decay generates the required lepton asymmetry. Domination of the energy budget of the Universe by the $\phi$ or the RHNs alters the evolution history of the primordial gravitational waves (PGW), of inflationary origin, which re-enter the horizon after inflation, modifying the spectral shape. The decays of $\phi$ and RHNs release entropy into the early Universe while nearly degenerate RHNs facilitate low and intermediate scale leptogenesis. We show that depending on the coupling $y_R$ of $\phi$ to radiation species, RHNs can achieve thermal abundance before decaying, which gives rise to thermal leptogenesis. A characteristic damping of the GW spectrum resulting in two knee-like features or one knee-like feature would provide evidence for low-scale thermal and non-thermal leptogenesis respectively. We explore the parameter space for the lightest right-handed neutrino mass $M_1\in[10^2,10^{14}]$ GeV and washout parameter $K$ that depends on the light-heavy neutrino Yukawa couplings $\lambda$, in the weak ($K < 1$) and strong ($K > 1$) washout regimes. The resulting novel features compatible with observed baryon asymmetry are detectable by future experiments like LISA and ET. By estimating signal-to-noise ratio (SNR) for upcoming GW experiments, we investigate the effect of the scalar mass $M_\phi$ and reheating temperature $T_\phi$, which depends on the $\phi-N$ Yukawa couplings $y_N$., Comment: 57 pages, 21 figures

Trans-Neptunian objects (TNOs) are small, icy bodies in the outer solar system. They are observed to have a complex orbital distribution that was shaped by the early dynamical history and migration of the giant planets. Comparisons between the different dynamical classes of modeled and observed TNOs can help constrain the history of the outer solar system. Because of the complex dynamics of TNOs, particularly those in and near mean motion resonances with Neptune, classification has traditionally been done by human inspection of plots of the time evolution of orbital parameters. This is very inefficient. The Vera Rubin Observatory's Legacy Survey of Space and Time (LSST) is expected to increase the number of known TNOs by a factor of $\sim$10, necessitating a much more automated process. In this chapter we present an improved supervised machine learning classifier for TNOs. Using a large and diverse training set as well as carefully chosen, dynamically motivated data features calculated from numerical integrations of TNO orbits, our classifier returns results that match those of a human classifier 98% of the time, and dynamically relevant classifications 99.7% of the time. This classifier is dramatically more efficient than human classification, and it will improve classification of both observed and modeled TNO data., Comment: Accepted chapter to appear in the Elsevier book "Machine Learning for Small Solar System Bodies", edited by Valerio Carruba, Evgeny Smirnov, and Dagmara Oszkiewicz

Extreme Multi-label Classification (XMC) involves predicting a subset of relevant labels from an extremely large label space, given an input query and labels with textual features. Models developed for this problem have conventionally used modular approach with (i) a Dual Encoder (DE) to embed the queries and label texts, (ii) a One-vs-All classifier to rerank the shortlisted labels mined through meta-classifier training. While such methods have shown empirical success, we observe two key uncharted aspects, (i) DE training typically uses only a single positive relation even for datasets which offer more, (ii) existing approaches fixate on using only OvA reduction of the multi-label problem. This work aims to explore these aspects by proposing UniDEC, a novel end-to-end trainable framework which trains the dual encoder and classifier in together in a unified fashion using a multi-class loss. For the choice of multi-class loss, the work proposes a novel pick-some-label (PSL) reduction of the multi-label problem with leverages multiple (in come cases, all) positives. The proposed framework achieves state-of-the-art results on a single GPU, while achieving on par results with respect to multi-GPU SOTA methods on various XML benchmark datasets, all while using 4-16x lesser compute and being practically scalable even beyond million label scale datasets.

We present a stability analysis of a large set of simulated planetary systems of three or more planets based on architectures of multiplanet systems discovered by \textit{Kepler} and \textit{K2}. We propagated 21,400 simulated planetary systems up to 5 billion orbits of the innermost planet; approximately 13% of these simulations ended in a planet-planet collision within that timespan. We examined trends in dynamical stability based on dynamical spacings, orbital period ratios, and mass ratios of nearest-neighbor planets as well as the system-wide planet mass distribution and the spectral fraction describing the system's short-term evolution. We find that instability is more likely in planetary systems with adjacent planet pairs that have period ratios less than two and in systems of greater variance of planet masses. Systems with planet pairs at very small dynamical spacings (less than $\sim10-12$ mutual Hill radius) are also prone to instabilities, but instabilities also occur at much larger planetary separations. We find that a large spectral fraction (calculated from short integrations) is a reasonable predictor of longer-term dynamical instability; systems that have a large number of Fourier components in their eccentricity vectors are prone to secular chaos and subsequent eccentricity growth and instabilities., Comment: accepted for publication in AJ

Multiple benchmarks have been developed to assess the alignment between deep neural networks (DNNs) and human vision. In almost all cases these benchmarks are observational in the sense they are composed of behavioural and brain responses to naturalistic images that have not been manipulated to test hypotheses regarding how DNNs or humans perceive and identify objects. Here we introduce the toolbox MindSet: Vision, consisting of a collection of image datasets and related scripts designed to test DNNs on 30 psychological findings. In all experimental conditions, the stimuli are systematically manipulated to test specific hypotheses regarding human visual perception and object recognition. In addition to providing pre-generated datasets of images, we provide code to regenerate these datasets, offering many configurable parameters which greatly extend the dataset versatility for different research contexts, and code to facilitate the testing of DNNs on these image datasets using three different methods (similarity judgments, out-of-distribution classification, and decoder method), accessible at https://github.com/MindSetVision/mindset-vision. We test ResNet-152 on each of these methods as an example of how the toolbox can be used.

In a recent publication (Toler et al. 2023), we demonstrated that for axisymmetric geometries, the Kapur-Rokhlin quadrature rule provided an efficient and high-order accurate method for computing the normal component, on the plasma surface, of the magnetic field due to the toroidal current flowing in the plasma, via the virtual-casing principle. The calculation was indirect, as it required the prior computation of the magnetic vector potential from the virtual-casing principle, followed by the computation of its tangential derivative by Fourier differentiation, in order to obtain the normal component of the magnetic field. Our approach did not provide the other components of the virtual-casing magnetic field. In this letter, we show that a more direct and more general approach is available for the computation of the virtual-casing magnetic field. The Kapur-Rokhlin quadrature rule accurately calculates the principal value integrals in the expression for all the components of the magnetic field on the plasma boundary, and the numerical error converges at a rate nearly as high as the indirect method we presented previously., Comment: 14 pages, 6 figures

Previous research underscored the potential of danmaku--a text-based commenting feature on videos--in engaging hearing audiences. Yet, for many Deaf and hard-of-hearing (DHH) individuals, American Sign Language (ASL) takes precedence over English. To improve inclusivity, we introduce "Signmaku," a new commenting mechanism that uses ASL, serving as a sign language counterpart to danmaku. Through a need-finding study (N=12) and a within-subject experiment (N=20), we evaluated three design styles: real human faces, cartoon-like figures, and robotic representations. The results showed that cartoon-like signmaku not only entertained but also encouraged participants to create and share ASL comments, with fewer privacy concerns compared to the other designs. Conversely, the robotic representations faced challenges in accurately depicting hand movements and facial expressions, resulting in higher cognitive demands on users. Signmaku featuring real human faces elicited the lowest cognitive load and was the most comprehensible among all three types. Our findings offered novel design implications for leveraging generative AI to create signmaku comments, enriching co-learning experiences for DHH individuals., Comment: 14 pages, CHI 2024

We study the generalization behavior of Markov Logic Networks (MLNs) across relational structures of different sizes. Multiple works have noticed that MLNs learned on a given domain generalize poorly across domains of different sizes. This behavior emerges from a lack of internal consistency within an MLN when used across different domain sizes. In this paper, we quantify this inconsistency and bound it in terms of the variance of the MLN parameters. The parameter variance also bounds the KL divergence between an MLN's marginal distributions taken from different domain sizes. We use these bounds to show that maximizing the data log-likelihood while simultaneously minimizing the parameter variance corresponds to two natural notions of generalization across domain sizes. Our theoretical results apply to Exponential Random Graphs and other Markov network based relational models. Finally, we observe that solutions known to decrease the variance of the MLN parameters, like regularization and Domain-Size Aware MLNs, increase the internal consistency of the MLNs. We empirically verify our results on four different datasets, with different methods to control parameter variance, showing that controlling parameter variance leads to better generalization., Comment: To Appear in Proceedings of ECML 2024-Research Track

Personas are commonly used in software projects to gain a better understanding of end-users' needs. However, there is a limited understanding of their usage and effectiveness in practice. This paper presents the results of a two-step investigation, comprising interviews with 26 software developers, UI/UX designers, business analysts and product managers and a survey of 203 practitioners, aimed at shedding light on the current practices, methods and challenges of using personas in software development. Our findings reveal variations in the frequency and effectiveness of personas across different software projects and IT companies, the challenges practitioners face when using personas and the reasons for not using them at all. Furthermore, we investigate the coverage of human aspects in personas, often assumed to be a key feature of persona descriptions. Contrary to the general perception, our study shows that human aspects are often ignored for various reasons in personas or requirements engineering in general. Our study provides actionable insights for practitioners to overcome challenges in using personas during requirements engineering stages, and we identify areas for future research.

We study the metallicity distribution and evolution in the Galactic disk based on the largest sample of open star clusters in the Galaxy. From the catalogue of 1879 open clusters in the range of Galactocentric distance (R_GC) from 4 to 20 kpc, we investigate the variation of metallicity in the Galactic disk as functions of R_GC, vertical distance (Z), and ages of the clusters. In the direction perpendicular to the Galactic plane, variation in metallicity is found to follow a stepped linear relation. We estimate a vertical metallicity gradient d[Fe/H]/dZ of -0.545+/-0.046 dex/kpc for |Z| < 0.487 kpc, and -0.075+/-0.093 dex/kpc for 0.487 < |Z| < 1.8 kpc. On average, metallicity variations above and below the Galactic plane are found to change at similar rates. The change in metallicity in the radial direction is also found to follow a two-function linear relation. We obtain a radial metallicity gradient d[Fe/H]/d[R_GC] of -0.070+/-0.002 dex/kpc for 4.0

We study a setting where an analyst has access to purely aggregate information about the consumption choices of a heterogenous population of individuals. We show that observing the statistical moments of market demand allows the analyst to test aggregate data for rationality. Interestingly, just the mean and variance of demand carry observable restrictions. This is in stark contrast to impossibility result of the Sonnenschein-Mantel-Debreu theorem, which shows that aggregate demand carries no observable restrictions at all. We leverage our approach to deliver a characterization of rationality in terms of moments for the common two-good case. We illustrate the usefulness of moment-based restrictions through two applications: (i) improving the precision of demand and welfare estimates; and (ii) testing for the existence of a welfare-relevant representative consumer., Comment: arXiv admin note: substantial text overlap with arXiv:2303.01231

This is a summary paper of a use case of a Robotdog dedicated to guide visually impaired people in complex environment like a smart intersection. In such scenarios, the Robotdog has to autonomously decide whether it is safe to cross the intersection or not in order to further guide the human. We leverage data sharing and collaboration between the Robotdog and other autonomous systems operating in the same environment. We propose a system architecture for autonomous systems through a separation of a collaborative decision layer, to enable collective decision making processes, where data about the environment, relevant to the Robotdog decision, together with evidences for trustworthiness about other systems and the environment are shared.

Recent Pulsar Timing Array (PTA) collaborations show strong evidence for a stochastic gravitational wave background (SGWB) with the characteristic Hellings-Downs inter-pulsar correlations. The signal may stem from supermassive black hole binary mergers, or early universe phenomena. The former is expected to be strongly anisotropic while primordial backgrounds are likely to be predominantly isotropic with small fluctuations. In case the observed SGWB is of cosmological origin, our relative motion with respect to the SGWB rest frame is a guaranteed source of anisotropy, leading to $\textit{O}(10^{-3})$ energy density fluctuations of the SGWB. For such cosmological SGWB, kinematic anisotropies are likely to be larger than the intrinsic anisotropies, akin to the cosmic microwave background (CMB) dipole anisotropy. We assess the sensitivity of current PTA data to the kinematic dipole anisotropy, and we also forecast at what extent the magnitude and direction of the kinematic dipole can be measured in the future with an SKA-like experiment. We also discuss how the spectral shape of the SGWB and the location of the pulsars to monitor affect the prospects of detecting the kinematic dipole with PTA. In the future, a detection of this anisotropy may even help resolve the discrepancy in the magnitude of the kinematic dipole as measured by CMB and large-scale structure observations., Comment: 29 pages, 6 figures. V2: corrected upper limit, updated forecasts

Deep Learning (DL) techniques have achieved remarkable successes in recent years. However, their ability to generalize and execute reasoning tasks remains a challenge. A potential solution to this issue is Neuro-Symbolic Integration (NeSy), where neural approaches are combined with symbolic reasoning. Most of these methods exploit a neural network to map perceptions to symbols and a logical reasoner to predict the output of the downstream task. These methods exhibit superior generalization capacity compared to fully neural architectures. However, they suffer from several issues, including slow convergence, learning difficulties with complex perception tasks, and convergence to local minima. This paper proposes a simple yet effective method to ameliorate these problems. The key idea involves pretraining a neural model on the downstream task. Then, a NeSy model is trained on the same task via transfer learning, where the weights of the perceptual part are injected from the pretrained network. The key observation of our work is that the neural network fails to generalize only at the level of the symbolic part while being perfectly capable of learning the mapping from perceptions to symbols. We have tested our training strategy on various SOTA NeSy methods and datasets, demonstrating consistent improvements in the aforementioned problems., Comment: Accepted as full paper at the International Conference on Neural-Symbolic Learning and Reasoning (NeSy 2024)