Search

Children spend most of their time at home in their early years, yet efforts to promote human capital at home in many low- and middle-income settings remain limited. We conduct a randomized controlled trial to evaluate an intervention which encourages parents and caregivers to foster human capital accumulation among their children between ages 3 and 5, with a focus on math and phonics skills. Children gain 0.52 and 0.51 standard deviations relative to the control group on math and phonics tests, respectively (p<0.001). A year later effects persist, but math gains dissipate to 0.15 (p=0.06) and phonics to 0.13 (p=0.12). Effects appear to be mediated largely through instructional support by parents and not other parent investment mechanisms, such as more positive parent-child interactions or additional time spent on education at home beyond the intervention. Our results show that parents can be effective conduits of educational instruction even in low-resource settings. [Funding for this paper was received from the Global Innovation Fund.]

Interference and scattering, often deemed undesirable, are inevitable in wireless communications, especially when the current mobile networks and upcoming sixth generation (6G) have turned into ultra-dense networks. Current approaches relying on multiple-input multiple-output (MIMO) combined with artificial-intelligence-aided (AI) signal processing have drawbacks of being power-hungry and requiring wide bandwidth that raise scalability concerns. In this article, we take a radical approach and utilize the channel fading phenomenon to our advantage. Specifically, we propose a novel meta-fluid antenna architecture, referred to as the `fluid' antenna system (FAS), that can freely surf on radio wave fluctuations, like `fluid' figuratively speaking, with fine resolution in space to opportunistically avoid interference, eliminating the need for expensive signal processing. Our experimental results demonstrate that under rich scattering conditions, the proposed meta-fluidic architecture is able to exploit the natural ups and downs of radio waves in space for spatial multiplexing. These breakthrough results show that scattering can be desirable not harmful and interference can be dodged not suppressed, fundamentally changing our perception of fading and our understanding on how interference should be managed in wireless communications networks.

Automated grading has become an essential tool in education technology due to its ability to efficiently assess large volumes of student work, provide consistent and unbiased evaluations, and deliver immediate feedback to enhance learning. However, current systems face significant limitations, including the need for large datasets in few-shot learning methods, a lack of personalized and actionable feedback, and an overemphasis on benchmark performance rather than student experience. To address these challenges, we propose a Zero-Shot Large Language Model (LLM)-Based Automated Assignment Grading (AAG) system. This framework leverages prompt engineering to evaluate both computational and explanatory student responses without requiring additional training or fine-tuning. The AAG system delivers tailored feedback that highlights individual strengths and areas for improvement, thereby enhancing student learning outcomes. Our study demonstrates the system's effectiveness through comprehensive evaluations, including survey responses from higher education students that indicate significant improvements in motivation, understanding, and preparedness compared to traditional grading methods. The results validate the AAG system's potential to transform educational assessment by prioritizing learning experiences and providing scalable, high-quality feedback.

Quick Merkle Database (QMDB) addresses longstanding bottlenecks in blockchain state management by integrating key-value (KV) and Merkle tree storage into a single unified architecture. QMDB delivers a significant throughput improvement over existing architectures, achieving up to 6X over the widely used RocksDB and 8X over NOMT, a leading verifiable database. Its novel append-only twig-based design enables one SSD read per state access, O(1) IOs for updates, and in-memory Merkleization on a memory footprint as small as 2.3 bytes per entry, enabling it to run on even modest consumer-grade PCs. QMDB scales seamlessly across both commodity and enterprise hardware, achieving up to 2.28 million state updates per second. This performance enables support for 1 million token transfers per second (TPS), marking QMDB as the first solution achieving such a milestone. QMDB has been benchmarked with workloads exceeding 15 billion entries (10X Ethereum's 2024 state) and has proven the capacity to scale to 280 billion entries on a single server. Furthermore, QMDB introduces historical proofs, unlocking the ability to query its blockchain's historical state at the latest block. QMDB not only meets the demands of current blockchains but also provides a robust foundation for building scalable, efficient, and verifiable decentralized applications across diverse use cases., Comment: 11 pages, 3 figures

Early detection of neurocognitive disorders (NCDs) is crucial for timely intervention and disease management. Speech analysis offers a non-intrusive and scalable screening method, particularly through narrative tasks in neuropsychological assessment tools. Traditional narrative analysis often focuses on local indicators in microstructure, such as word usage and syntax. While these features provide insights into language production abilities, they often fail to capture global narrative patterns, or microstructures. Macrostructures include coherence, thematic organization, and logical progressions, reflecting essential cognitive skills potentially critical for recognizing NCDs. Addressing this gap, we propose to investigate specific cognitive and linguistic challenges by analyzing topical shifts, temporal dynamics, and the coherence of narratives over time, aiming to reveal cognitive deficits by identifying narrative impairments, and exploring their impact on communication and cognition. The investigation is based on the CU-MARVEL Rabbit Story corpus, which comprises recordings of a story-telling task from 758 older adults. We developed two approaches: the Dynamic Topic Models (DTM)-based temporal analysis to examine the evolution of topics over time, and the Text-Image Temporal Alignment Network (TITAN) to evaluate the coherence between spoken narratives and visual stimuli. DTM-based approach validated the effectiveness of dynamic topic consistency as a macrostructural metric (F1=0.61, AUC=0.78). The TITAN approach achieved the highest performance (F1=0.72, AUC=0.81), surpassing established microstructural and macrostructural feature sets. Cross-comparison and regression tasks further demonstrated the effectiveness of proposed dynamic macrostructural modeling approaches for NCD detection., Comment: 12 pages, 8 figures

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 this paper, we investigate the ability of single-layer attention-only transformers (i.e. attention layers) to memorize facts contained in databases from a linear-algebraic perspective. We associate with each database a 3-tensor, propose the rank of this tensor as a measure of the size of the database, and provide bounds on the rank in terms of properties of the database. We also define a 3-tensor corresponding to an attention layer, and empirically demonstrate the relationship between its rank and database rank on a dataset of toy models and random databases. By highlighting the roles played by the value-output and query-key weights, and the effects of argmax and softmax on rank, our results shed light on the `additive motif' of factual recall in transformers, while also suggesting a way of increasing layer capacity without increasing the number of parameters.

Ultracold quantum-gas mixtures of fermionic atoms with resonant control of interactions offer a unique test-bed to explore few- and many-body quantum states with unconventional properties. The emergence of such strongly correlated systems, as for instance symmetry-broken superfluids, is usually accompanied by hydrodynamic collective behavior. Thus, experimental progress in this field naturally requires a deep understanding of hydrodynamic regimes. Here, we report on experiments employing a tunable Fermi-Fermi mixture of $^{161}$Dy and $^{40}$K near quantum degeneracy. We investigate the full spectrum of dipole modes across a Feshbach resonance and characterize the crossover from collisionless to deep hydrodynamic behavior in measurements of frequencies and damping rates. We compare our results with a theoretical model that considers the motion of the mass centers of the two species and we identify the contributions of friction and mean-field interaction. We show that one oscillating mode exists over the whole range of interactions, exhibiting striking changes of frequency and damping in the deep hydrodynamic regime. We observe the second oscillating mode to split into two purely exponential damping modes. One of these exponential modes shows very fast damping, faster than any other relevant timescale, and is largely insensitive against experimental imperfections. It provides an accurate measure for the interspecies drag effect, which generalizes the concept of spin drag explored in other experiments. We characterize the interspecies drag locally in terms of a microscopic friction coefficient and we discuss its unitarity-limited universal behavior on top of the resonance., Comment: 20 pages, 11 figures

Visual storytelling combines visuals and narratives to communicate important insights. While web-based visual storytelling is well-established, leveraging the next generation of digital technologies for visual storytelling, specifically immersive technologies, remains underexplored. We investigated the impact of the story viewpoint (from the audience's perspective) and navigation (when progressing through the story) on spatial immersion and understanding. First, we collected web-based 3D stories and elicited design considerations from three VR developers. We then adapted four selected web-based stories to an immersive format. Finally, we conducted a user study (N=24) to examine egocentric and exocentric viewpoints, active and passive navigation, and the combinations they form. Our results indicated significantly higher preferences for egocentric+active (higher agency and engagement) and exocentric+passive (higher focus on content). We also found a marginal significance of viewpoints on story understanding and a strong significance of navigation on spatial immersion.

Generating synthetic datasets via large language models (LLMs) themselves has emerged as a promising approach to improve LLM performance. However, LLMs inherently reflect biases present in their training data, leading to a critical challenge: when these models generate synthetic data for training, they may propagate and amplify their inherent biases that can significantly impact model fairness and robustness on downstream tasks--a phenomenon we term bias inheritance. This work presents the first systematic investigation in understanding, analyzing, and mitigating bias inheritance. We study this problem by fine-tuning LLMs with a combined dataset consisting of original and LLM-augmented data, where bias ratio represents the proportion of augmented data. Through systematic experiments across 10 classification and generation tasks, we analyze how 6 different types of biases manifest at varying bias ratios. Our results reveal that bias inheritance has nuanced effects on downstream tasks, influencing both classification tasks and generation tasks differently. Then, our analysis identifies three key misalignment factors: misalignment of values, group data, and data distributions. Based on these insights, we propose three mitigation strategies: token-based, mask-based, and loss-based approaches. Experiments demonstrate that these strategies also work differently on various tasks and bias, indicating the substantial challenges to fully mitigate bias inheritance. We hope this work can provide valuable insights to the research of LLM data augmentation., Comment: Technical report; 31 pages

This paper presents a method that allows users to design cinematic video shots in the context of image-to-video generation. Shot design, a critical aspect of filmmaking, involves meticulously planning both camera movements and object motions in a scene. However, enabling intuitive shot design in modern image-to-video generation systems presents two main challenges: first, effectively capturing user intentions on the motion design, where both camera movements and scene-space object motions must be specified jointly; and second, representing motion information that can be effectively utilized by a video diffusion model to synthesize the image animations. To address these challenges, we introduce MotionCanvas, a method that integrates user-driven controls into image-to-video (I2V) generation models, allowing users to control both object and camera motions in a scene-aware manner. By connecting insights from classical computer graphics and contemporary video generation techniques, we demonstrate the ability to achieve 3D-aware motion control in I2V synthesis without requiring costly 3D-related training data. MotionCanvas enables users to intuitively depict scene-space motion intentions, and translates them into spatiotemporal motion-conditioning signals for video diffusion models. We demonstrate the effectiveness of our method on a wide range of real-world image content and shot-design scenarios, highlighting its potential to enhance the creative workflows in digital content creation and adapt to various image and video editing applications., Comment: It is best viewed in Acrobat. Project page: https://motion-canvas25.github.io/

An emerging fluid antenna system (FAS) brings a new dimension, i.e., the antenna positions, to deal with the deep fading, but simultaneously introduces challenges related to the transmit design. This paper proposes an ``unsupervised learning to optimize" paradigm to optimize the FAS. Particularly, we formulate the sum-rate and energy efficiency (EE) maximization problems for a multiple-user multiple-input single-output (MU-MISO) FAS and solved by a two-stage graph neural network (GNN) where the first stage and the second stage are for the inference of antenna positions and beamforming vectors, respectively. The outputs of the two stages are jointly input into a unsupervised loss function to train the two-stage GNN. The numerical results demonstrates that the advantages of the FAS for performance improvement and the two-stage GNN for real-time and scalable optimization. Besides, the two stages can function separately.

While the sources of the diffuse astrophysical neutrino flux detected by the IceCube Neutrino Observatory are still largely unknown, one of the promising methods used towards understanding this is investigating the potential temporal and spatial correlations between neutrino alerts and the electromagnetic radiation from blazars. We report on the multiwavelength target-of-opportunity observations of the blazar B3 2247+381, taken in response to an IceCube multiplet alert for a cluster of muon neutrino events compatible with the source location between May 20, 2022 and November 10, 2022. B3 2247+381 was not detected with VERITAS during this time period. The source was found to be in a low-flux state in the optical, ultraviolet and gamma-ray bands for the time interval corresponding to the neutrino event, but was detected in the hard X-ray band with NuSTAR during this period. We find the multiwavelength spectral energy distribution is well described using a simple one-zone leptonic synchrotron self-Compton radiation model. Moreover, assuming the neutrinos originate from hadronic processes within the jet, the neutrino flux would be accompanied by a photon flux from the cascade emission, and the integrated photon flux required in such a case would significantly exceed the total multiwavelength fluxes and the VERITAS upper limits presented here. The lack of flaring activity observed with VERITAS, combined with the low multiwavelength flux levels, and given the significance of the neutrino excess is at 3$\sigma$ level (uncorrected for trials), makes B3 2247+381 an unlikely source of the IceCube multiplet. We conclude that the neutrino excess is likely a background fluctuation., Comment: 26 pages, 5 figures. Accepted for publication in the Astrophysical Journal (ApJ)

The presence of GPU from different vendors demands the Lattice QCD codes to support multiple architectures. To this end, Open Computing Language (OpenCL) is one of the viable frameworks for writing a portable code. It is of interest to find out how the OpenCL implementation performs as compared to the code based on a dedicated programming interface such as CUDA for Nvidia GPUs. We have developed an OpenCL backend for our already existing code of the Wuppertal-Budapest collaboration. In this contribution, we show benchmarks of the most time consuming part of the numerical simulation, namely, the inversion of the Dirac operator. We present the code performance on the JUWELS and LUMI Supercomputers based on Nvidia and AMD graphics cards, respectively, and compare with the CUDA backend implementation., Comment: Proceedings for Lattice Conference 2024

Near a critical endpoint the Lee-Yang edge singularity approaches the real axis in the complex chemical potential plane. In the vicinity of the critical point the functional form of this approach depends on the universality class. Assuming a three dimensional Ising critical point in the QCD phase diagram the location of the critical endpoint can be extrapolated provided that the position of the Lee-Yang edge singularity is known at multiple temperatures. A popular method to estimate the position of a singularity is to model the free energy as a rational function of the baryon chemical potential $\mu_\text{B}$. The parameters of this model can be constrained by the cumulants of the net baryon density taken at $\mu_\text{B}^2\leq0$ . Using high-statistics simulations on a lattice $16^3\times8$ by the Wuppertal-Budapest Collaboration we estimate the location of the closest singularity in the QCD phase diagram. We also compare various models for the functional form of the free energy and discuss the predictive power of this approach., Comment: Proceedings to Lattice Conference 2024

Turbulent, relativistic nonthermal plasmas are ubiquitous in high-energy astrophysical systems, as inferred from broadband nonthermal emission spectra. The underlying turbulent nonthermal particle acceleration (NTPA) processes have traditionally been modelled with a Fokker-Planck (FP) diffusion-advection equation for the particle energy distribution. We test FP-type NTPA theories by performing and analysing particle-in-cell (PIC) simulations of turbulence in collisionless relativistic pair plasma. By tracking large numbers of particles in simulations with different initial magnetisation and system size, we first test and confirm the applicability of the FP framework. We then measure the FP energy diffusion ($D$) and advection ($A$) coefficients as functions of particle energy $\gamma m c^2$, and compare their dependence to theoretical predictions. At high energies, we robustly find $D \sim \gamma^2$ for all cases. Hence, we fit $D = D_0 \gamma^2$ and find a scaling consistent with $D_0 \sim \sigma^{3/2}$ at low instantaneous magnetisation $\sigma(t)$, flattening to $D_0 \sim \sigma$ at higher $\sigma \sim 1$. We also find that the power-law index $\alpha(t)$ of the particle energy distribution converges exponentially in time. We build and test an analytic model connecting the FP coefficients and $\alpha(t)$, predicting $A(\gamma) \sim \gamma \log \gamma$. We confirm this functional form in our measurements of $A(\gamma,t)$, which allows us to predict $\alpha(t)$ through the model relations. Our results suggest that the basic second-order Fermi acceleration model, which predicts $D_0 \sim \sigma$, may not be a complete description of NTPA in turbulent plasmas. These findings encourage further application of tracked particles and FP coefficients as a diagnostic in kinetic simulations of various astrophysically relevant plasma processes like collisionless shocks and magnetic reconnection., Comment: 22 pages, 24 figures, submitted for publication. Comments are welcome!

The contact issue for two-dimensional (2D) materials-based field-effect transistors (FETs) has drawn enormous attention in recent years. Although ohmic behavior is achieved at room temperature, the drain current of 2DFETs shifts from ohmic to non-ohmic behavior at cryogenic temperatures. In this work, we demonstrate that the shift is attributed to the asymmetric current reduction at the metal-semiconductor contact at low temperature. Under low drain bias, carriers tunnel from the source to the channel but diffuse to the drain side due to the channel-to-drain barrier, resulting in the current suppression. By studying the property of ohmic metal-semiconductor contact at different temperatures, we analyzed the mechanisms behind this phenomenon and the dependence on metal-to-semiconductor barrier height. The work opens the semiconductor physics of 2D material contact at cryogenic temperature and the importance of contact metal selection in the development of 2DFET at cryogenic temperature., Comment: 13 pages, 5 figures

Stance detection has emerged as a popular task in natural language processing research, enabled largely by the abundance of target-specific social media data. While there has been considerable research on the development of stance detection models, datasets, and application, we highlight important gaps pertaining to (i) a lack of theoretical conceptualization of stance, and (ii) the treatment of stance at an individual- or user-level, as opposed to message-level. In this paper, we first review the interdisciplinary origins of stance as an individual-level construct to highlight relevant attributes (e.g., psychological features) that might be useful to incorporate in stance detection models. Further, we argue that recent pre-trained and large language models (LLMs) might offer a way to flexibly infer such user-level attributes and/or incorporate them in modelling stance. To better illustrate this, we briefly review and synthesize the emerging corpus of studies on using LLMs for inferring stance, and specifically on incorporating user attributes in such tasks. We conclude by proposing a four-point agenda for pursuing stance detection research that is theoretically informed, inclusive, and practically impactful.

The vast majority of real-world patient information resides in unstructured clinical text, and the process of medical abstraction seeks to extract and normalize structured information from this unstructured input. However, traditional medical abstraction methods can require significant manual efforts that can include crafting rules or annotating training labels, limiting scalability. In this paper, we propose UniMedAbstractor (UMA), a zero-shot medical abstraction framework leveraging Large Language Models (LLMs) through a modular and customizable prompt template. We refer to our approach as universal abstraction as it can quickly scale to new attributes through its universal prompt template without curating attribute-specific training labels or rules. We evaluate UMA for oncology applications, focusing on fifteen key attributes representing the cancer patient journey, from short-context attributes (e.g., performance status, treatment) to complex long-context attributes requiring longitudinal reasoning (e.g., tumor site, histology, TNM staging). Experiments on real-world data show UMA's strong performance and generalizability. Compared to supervised and heuristic baselines, UMA with GPT-4o achieves on average an absolute 2-point F1/accuracy improvement for both short-context and long-context attribute abstraction. For pathologic T staging, UMA even outperforms the supervised model by 20 points in accuracy.

We built a spatial hybrid system that combines a personal computer (PC) and virtual reality (VR) for visual sensemaking, addressing limitations in both environments. Although VR offers immense potential for interactive data visualization (e.g., large display space and spatial navigation), it can also present challenges such as imprecise interactions and user fatigue. At the same time, a PC offers precise and familiar interactions but has limited display space and interaction modality. Therefore, we iteratively designed a spatial hybrid system (PC+VR) to complement these two environments by enabling seamless switching between PC and VR environments. To evaluate the system's effectiveness and user experience, we compared it to using a single computing environment (i.e., PC-only and VR-only). Our study results (N=18) showed that spatial PC+VR could combine the benefits of both devices to outperform user preference for VR-only without a negative impact on performance from device switching overhead. Finally, we discussed future design implications., Comment: Accepted by IEEE TVCG

The combinatorial multi-armed bandit (CMAB) is a fundamental sequential decision-making framework, extensively studied over the past decade. However, existing work primarily focuses on the online setting, overlooking the substantial costs of online interactions and the readily available offline datasets. To overcome these limitations, we introduce Off-CMAB, the first offline learning framework for CMAB. Central to our framework is the combinatorial lower confidence bound (CLCB) algorithm, which combines pessimistic reward estimations with combinatorial solvers. To characterize the quality of offline datasets, we propose two novel data coverage conditions and prove that, under these conditions, CLCB achieves a near-optimal suboptimality gap, matching the theoretical lower bound up to a logarithmic factor. We validate Off-CMAB through practical applications, including learning to rank, large language model (LLM) caching, and social influence maximization, showing its ability to handle nonlinear reward functions, general feedback models, and out-of-distribution action samples that excludes optimal or even feasible actions. Extensive experiments on synthetic and real-world datasets further highlight the superior performance of CLCB.

The electric dipole moments~(EDM) of fundamental particles inherently violate parity~(P) and time-reversal~(T) symmetries. By virtue of the CPT theorem in quantum field theory, the latter also implies the violation of the combined charge-conjugation and parity~(CP) symmetry. We aim to measure the EDM of the muon using the frozen-spin technique within a compact storage trap. This method exploits the high effective electric field, \$E \approx 165\$ MV/m, experienced in the rest frame of the muon with a momentum of about 23 MeV/c when it passes through a solenoidal magnetic field of \$|\vec{B}|=2.5\$ T. In this paper, we outline the fundamental considerations for a muon EDM search and present a conceptual design for a demonstration experiment to be conducted at secondary muon beamlines of the Paul Scherrer Institute in Switzerland. In Phase~I, with an anticipated data acquisition period of 200 days, the expected sensitivity to a muon EDM is 4E-21 ecm. In a subsequent phase, Phase~II, we propose to improve the sensitivity to 6E-23 ecm using a dedicated instrument installed on a different beamline that produces muons of momentum 125 MeV/c}., Comment: 34 pages, submitted to EPJC

Air pollution (AP) poses a great threat to human health, and people are paying more attention than ever to its prediction. Accurate prediction of AP helps people to plan for their outdoor activities and aids protecting human health. In this paper, long-short term memory (LSTM) recurrent neural networks (RNNs) have been used to predict the future concentration of air pollutants (APS) in Macau. Additionally, meteorological data and data on the concentration of APS have been utilized. Moreover, in Macau, some air quality monitoring stations (AQMSs) have less observed data in quantity, and, at the same time, some AQMSs recorded less observed data of certain types of APS. Therefore, the transfer learning and pre-trained neural networks have been employed to assist AQMSs with less observed data to build a neural network with high prediction accuracy. The experimental sample covers a period longer than 12-year and includes daily measurements from several APS as well as other more classical meteorological values. Records from five stations, four out of them are AQMSs and the remaining one is an automatic weather station, have been prepared from the aforesaid period and eventually underwent to computational intelligence techniques to build and extract a prediction knowledge-based system. As shown by experimentation, LSTM RNNs initialized with transfer learning methods have higher prediction accuracy; it incurred shorter training time than randomly initialized recurrent neural networks., Comment: Forecasting, environment monitoring, transfer learning, recurrent neural network, airborne particle matter

We demonstrate fluorescence generation through a sequential multiphoton process with real intermediary states. Our findings suggest new directions for optical control of previously unexplored molecular excitation pathways.

Object removal is of great significance to 3D scene understanding, essential for applications in content filtering and scene editing. Current mainstream methods primarily focus on removing individual objects, with a few methods dedicated to eliminating an entire area or all objects of a certain category. They however confront the challenge of insufficient granularity and flexibility for real-world applications, where users demand tailored excision and preservation of objects within defined zones. In addition, most of the current methods require kinds of priors when addressing multi-view inpainting, which is time-consuming. To address these limitations, we propose an efficient and user-friendly pipeline for 3D multi-object removal, enabling users to flexibly select areas and define objects for removal or preservation. Concretely, to ensure object consistency and correspondence across multiple views, we propose a novel mask matching and refinement module, which integrates homography-based warping with high-confidence anchor points for segmentation. By leveraging the IoU joint shape context distance loss, we enhance the accuracy of warped masks and improve subsequent inpainting processes. Considering the current immaturity of 3D multi-object removal, we provide a new evaluation dataset to bridge the developmental void. Experimental results demonstrate that our method significantly reduces computational costs, achieving processing speeds more than 80% faster than state-of-the-art methods while maintaining equivalent or higher reconstruction quality.

We aim to compute the impact rates for objects with a diameter of 1 km onto the regular satellites of Jupiter, Saturn and Uranus using our latest dynamical simulations of the evolution of outer solar system coupled with the best estimates of the current population of objects beyond Neptune and their size-frequency distribution. We use the outcome of the last 3.5~Gyr of evolution of the outer solar system from our database of simulations and combine this with observational constraints of the population beyond Neptune to compute the flux of objects entering the Centaur region, with uncertainties. The initial conditions resemble the current population rather than a near-circular, near-planar disc usually assumed just before the onset of giant planet migration. We obtain a better estimate of the impact probability of a Centaur with the satellites from enacting simulations of planetesimals flying past the satellites on hyperbolic orbits, which agree with literature precedents. We find that our impact rate of objects greater than 1 km in diameter with Jupiter is 0.0012/yr, which is a factor of 3--6 lower than previous estimates of 0.0044/yr from Nesvorny et al. (2023) and 0.0075/yr from Zahnle et al. (2003). On the other hand our impact probabilities with the satellites scaled to the giant planets are consistent with these earlier literature estimates, as is the leakage rate of objects from beyond Neptune into the Centaur region. However, our absolute impact probabilities with the giant planets are lower. We attribute this to our choice of initial conditions., Comment: In revision with Astronomy and Astrophysics

The first International AI Safety Report comprehensively synthesizes the current evidence on the capabilities, risks, and safety of advanced AI systems. The report was mandated by the nations attending the AI Safety Summit in Bletchley, UK. Thirty nations, the UN, the OECD, and the EU each nominated a representative to the report's Expert Advisory Panel. A total of 100 AI experts contributed, representing diverse perspectives and disciplines. Led by the report's Chair, these independent experts collectively had full discretion over the report's content.

Amorphous oxide semiconductors are gaining interest for logic and memory transistors compatible with low-temperature fabrication. However, their low thermal conductivity and heterogeneous interfaces suggest that their performance may be severely limited by self-heating, especially at higher power and device densities. Here, we investigate the high-field breakdown of amorphous indium tin oxide (ITO) transistors with scanning thermal microscopy (SThM) and multiphysics simulations. The ITO devices break irreversibly at channel temperatures of ~180 {\deg}C and ~340 {\deg}C on SiO${_2}$ and HfO${_2}$ substrates, respectively, but failure appears primarily caused by thermally-induced compressive strain near the device contacts. Combining SThM measurements with simulations allows us to estimate a thermal boundary conductance (TBC) of 35 ${\pm}$ 12 MWm${^-}$${^2}$K${^-}$${^1}$ for ITO on SiO${_2}$, and 51 ${\pm}$ 14 MWm${^-}$${^2}$K${^-}$${^1}$ for ITO on HfO${_2}$. The latter also enables significantly higher breakdown power due to better heat dissipation and closer thermal expansion matching. These findings provide valuable insights into the thermo-mechanical limitations of ITO devices, paving the way for more reliable and high-performance amorphous oxide transistors.

We present the third phase of the largest high-frequency, high-resolution imaging survey of 231 nearby, hard X-ray selected AGN, with a very high $98 \pm 1\%$ detection fraction. This survey presents VLA 22 GHz radio observations with 1" spatial resolution covering over $6$ orders of magnitude in radio luminosity in nearby AGN that span $\sim4$ orders of magnitude in black hole mass and X-ray luminosity. We identify three different radio morphologies: $44 \pm 3\%$ (102/231) are compact or unresolved, $46 \pm 3\%$ (106/231) show an extended structure (star formation, possible one-sided jets, etc.), and $8 \pm 2\%$ (19/231) have a biconical or two-sided jet-like morphology. The remaining $2 \pm 1\%$ (4/231) sources are non-detections. The radio-to-X-ray luminosity ratios of the Swift-BAT AGN ($\text{L}_R/\text{L}_{14-195 \text{keV}} \sim 10^{-5.5}$ and $\text{L}_R/\text{L}_{2-10 \text{keV}} \sim 10^{-5}$) with a scatter of $\sim0.5$ dex are similar to that of coronally active stars ($\text{L}_R/\text{L}_X \sim 10^{-5}$). For most targets, extended emission in radio-quiet objects is broadly consistent with the expectation for star formation from previous FIR observations, once the contribution from the radio core has been subtracted. Our sample represents nearby analogs of distant AGN at the peak of black hole growth, and thus the high detection fraction in our work has important implications for future high frequency AGN radio surveys with the next generation VLA (ngVLA) or Square Kilometre Array (SKA), both of which should detect large fractions of more distant AGN., Comment: 26 pages, 8 figures, 4tables. Accepted for publication in ApJ

Near-earth asteroid (NEA) 1998 KY$_{26}$ is a target of the $Hayabusa2\#$ spacecraft, which it will rendezvous with in July 2031. The asteroid has been noted to rotate rapidly and has a large out-of-plane nongravitational acceleration. We present observations consisting of deep g and R band imaging obtained with the Keck I/Low Resolution Imaging Spectrometer (LRIS) and visible spectroscopy from Gemini North /Gemini Multi-Object Spectrograph (Gemini N/GMOS) taken of 1998 KY$_{26}$ on 2024 June 8-9 when the asteroid was $\sim$0.037 au from the Earth. The asteroid lacks the appearance of a dust coma and has a surface brightness profile similar to nearby background stars in the deep images. The spectrum of 1998 KY$_{26}$ from the combined LRIS/GMOS observations most closely resembles Xe-types possessing a spectral slope of 6.71$\pm$0.43 $\%$ 100 nm$^{-1}$, and color indices: g-r = 0.63$\pm$0.03, r-i = 0.15$\pm$0.03, i-z = 0.05$\pm$0.04. From our deep image stacks, we compute a 3$\sigma$ upper limit on the dust production of 1998 KY$_{26}$ of $<$10$^{-5}$ kg s$^{-1}$, $<$10$^{-2}$ kg s$^{-1}$, and $<$10$^{-1}$ kg s$^{-1}$ assuming $\mathrm{\mu}$m, mm, and cm size dust particles. In addition, we compare the orbit of 1998 KY$_{26}$ and other known asteroids with large nongravitational parameters with NEA population models, finding the majority, including 1998 KY$_{26}$, likely originated from the inner Main Belt, while the second most numerous group originates from the outer Main Belt, followed by a third group possibly originating from the Jupiter Family Comet population. Given its inner Main Belt origin, its Xe-type, and rapid rotation, we hypothesize that the nongravitational acceleration of 1998 KY$_{26}$ may be caused by the shedding of large dust grains from its surface due to its rotation rather than H$_2$O vapor outgassing., Comment: 12 pages, 2 figures, 1 table

Continuous-time quantum walks are typically effected by either the discrete Laplacian or the adjacency matrix. In this paper, we explore a third option: the signless Laplacian, which has applications in algebraic graph theory and may arise in layered antiferromagnetic materials. We explore spatial search on the complete bipartite graph, which is generally irregular and breaks the equivalence of the three quantum walks for regular graphs, and where the search oracle breaks the equivalence of the Laplacian and signless Laplacian quantum walks on bipartite graphs without the oracle. We prove that a uniform superposition over all the vertices of the graph partially evolves to the marked vertices in one partite set, with the choice of set depending on the jumping rate of the quantum walk. We boost this success probability to 1 by proving that a particular non-uniform initial state completely evolves to the marked vertices in one partite set, again depending on the jumping rate. For some parameter regimes, the signless Laplacian yields the fastest search algorithm of the three, suggesting that it could be a new tool for developing faster quantum algorithms., Comment: 15 pages, 6 figures

Photonic crystal fibres (PCFs) are optical fibres that guide light using a modulated dielectric medium. They provide an exceptionally versatile platform for various applications, thanks to the flexibility with which light-guiding can be customised by modifying the fibre geometry. Here, we realise a PCF with guided modes produced by photonic bandstructure topology rather than conventional mode-trapping mechanisms. The design, which is compatible with the stack-and-draw fabrication process, consists of a cross-sectional photonic topological crystalline insulator with a disclination. A bulk-defect correspondence produces degenerate topological modes, lying below the cladding light line. We use various theoretical methods to confirm their topological origins, including a spectral localiser that makes minimal assumptions about the bandstructure. Our experiments on the fabricated topological fibre show it transmitting visible to near-infrared light with low losses of 10--20 dB/km, which do not increase much when the fibre is bent. A comparable solid-core PCF of conventional design exhibits substantially higher bending losses. Optical fibres based on topological modes thus hold promise for improved performance and novel functionalities.

Latent space matching, which consists of matching distributions of features in latent space, is a crucial component for tasks such as adversarial attacks and defenses, domain adaptation, and generative modelling. Metrics for probability measures, such as Wasserstein and maximum mean discrepancy, are commonly used to quantify the differences between such distributions. However, these are often costly to compute, or do not appropriately take the geometric and topological features of the distributions into consideration. Persistent homology is a tool from topological data analysis which quantifies the multi-scale topological structure of point clouds, and has recently been used as a topological regularizer in learning tasks. However, computation costs preclude larger scale computations, and discontinuities in the gradient lead to unstable training behavior such as in adversarial tasks. We propose the use of principal persistence measures, based on computing the persistent homology of a large number of small subsamples, as a topological regularizer. We provide a parallelized GPU implementation of this regularizer, and prove that gradients are continuous for smooth densities. Furthermore, we demonstrate the efficacy of this regularizer on shape matching, image generation, and semi-supervised learning tasks, opening the door towards a scalable regularizer for topological features., Comment: 31 pages, accepted to ICLR 2025

For a positive integer $m$, a finite group $G$ is said to admit a tournament $m$-semiregular representation (TmSR for short) if there exists a tournament $\Gamma$ such that the automorphism group of $\Gamma$ is isomorphic to $G$ and acts semiregularly on the vertex set of $\Gamma$ with $m$ orbits. Clearly, every finite group of even order does not admit a TmSR for any positive integer $m$, and T1SR is the well-known tournament regular representation (TRR for short). In 1986, Godsil \cite{god} proved, by a probabilistic approach, that the only finite groups of odd order without a TRR are $\mathbb{Z}_3^2$ and $\mathbb{Z}_3^3$ . More recently, Du \cite{du} proved that every finite group of odd order has a TmSR for every $m \geq 2$. The author of \cite{du} observed that a finite group of odd order has no regular TmSR when $m$ is an even integer, a group of order $1$ has no regular T3SR, and $\mathbb{Z}_3^2$ admits a regular T3SR. At the end of \cite{du}, Du proposed the following problem. \noindent{\sf\it Problem.} \ \ {\it For every odd integer $m\geq 3$, classify finite groups of odd order which have a regular TmSR.} The motivation of this paper is to give an answer for the above problem. We proved that if $G$ is a finite group with odd order $n>1$, then $G$ admits a regular TmSR for any odd integer $m\geq 3$.

We study the problem of computing the Fr\'echet distance between two polygonal curves under transformations. First, we consider translations in the Euclidean plane. Given two curves $\pi$ and $\sigma$ of total complexity $n$ and a threshold $\delta \geq 0$, we present an $\tilde{\mathcal{O}}(n^{7 + \frac{1}{3}})$ time algorithm to determine whether there exists a translation $t \in \mathbb{R}^2$ such that the Fr\'echet distance between $\pi$ and $\sigma + t$ is at most $\delta$. This improves on the previous best result, which is an $\mathcal{O}(n^8)$ time algorithm. We then generalize this result to any class of rationally parameterized transformations, which includes translation, rotation, scaling, and arbitrary affine transformations. For a class $\mathcal T$ of rationally parametrized transformations with $k$ degrees of freedom, we show that one can determine whether there is a transformation $\tau \in \mathcal T$ such that the Fr\'echet distance between $\pi$ and $\tau(\sigma)$ is at most $\delta$ in $\tilde{\mathcal{O}}(n^{3k+\frac{4}{3}})$ time., Comment: under submission

Recent video inpainting methods have achieved encouraging improvements by leveraging optical flow to guide pixel propagation from reference frames either in the image space or feature space. However, they would produce severe artifacts in the mask center when the masked area is too large and no pixel correspondences can be found for the center. Recently, diffusion models have demonstrated impressive performance in generating diverse and high-quality images, and have been exploited in a number of works for image inpainting. These methods, however, cannot be applied directly to videos to produce temporal-coherent inpainting results. In this paper, we propose a training-free framework, named VipDiff, for conditioning diffusion model on the reverse diffusion process to produce temporal-coherent inpainting results without requiring any training data or fine-tuning the pre-trained diffusion models. VipDiff takes optical flow as guidance to extract valid pixels from reference frames to serve as constraints in optimizing the randomly sampled Gaussian noise, and uses the generated results for further pixel propagation and conditional generation. VipDiff also allows for generating diverse video inpainting results over different sampled noise. Experiments demonstrate that VipDiff can largely outperform state-of-the-art video inpainting methods in terms of both spatial-temporal coherence and fidelity., Comment: 10 pages, 5 Figures (Accepted at WACV 2025)

Flexible-antenna systems, such as fluid antennas and movable antennas, have been recognized as key enabling technologies for sixth-generation (6G) wireless networks, as they can intelligently reconfigure the effective channel gains of the users and hence significantly improve their data transmission capabilities. However, existing flexible-antenna systems have been designed to combat small-scale fading in non-line-of-sight (NLoS) conditions. As a result, they lack the ability to establish line-of-sight links, which are typically 100 times stronger than NLoS links. In addition, existing flexible-antenna systems have limited flexibility, where adding/removing an antenna is not straightforward. This article introduces an innovative flexible-antenna system called pinching antennas, which are realized by applying small dielectric particles to waveguides. We first describe the basics of pinching-antenna systems and their ability to provide strong LoS links by deploying pinching antennas close to the users as well as their capability to scale up/down the antenna system. We then focus on communication scenarios with different numbers of waveguides and pinching antennas, where innovative approaches to implement multiple-input multiple-output and non-orthogonal multiple access are discussed. In addition, promising 6G-related applications of pinching antennas, including integrated sensing and communication and next-generation multiple access, are presented. Finally, important directions for future research, such as waveguide deployment and channel estimation, are highlighted.

We consider a generalization of the Kauffman bracket skein algebra of a surface that is generated by loops and arcs between marked points on the interior or boundary, up to skein relations defined by Muller and Roger-Yang. We compute the center of this Muller-Roger-Yang skein algebra and show that it is almost Azumaya when the quantum parameter $q$ is a primitive $n$-th root of unity with odd $n$. We also discuss the implications on the representation theory of the Muller-Roger-Yang generalized skein algebra., Comment: 28 pages, 10 figures

The Electron-Ion Collider (EIC) is an upcoming accelerator facility aimed at exploring the properties of quarks and gluons in nucleons and nuclei, shedding light on their structure and dynamics. The inaugural experimental apparatus, ePIC (electron-Proton and Ion Collider), is designed as a general purpose detector to address the NAS/NSAC physics program at the EIC. The wider EIC community is strongly supporting a second interaction region and associated second detector to enhance the full science program. In this study, we evaluate how the second interaction region and detector can be complementary to ePIC. The pre-conceptual layout of an interaction region for the second detector offers a secondary focus that provides better forward detector acceptance at scattering angles near $\theta \sim 0$~mrad, which can specifically enhance the exclusive, tagging, and diffractive physics program. This article presents an analysis of a tagging program using the second interaction region layout with incoherent diffractive vector meson production. The potential for the second interaction region to provide improved vetoing capabilities for incoherent events to elucidate the coherent diffractive cross-section is evaluated. The capability to access the coherent diffractive cross-section is of prime importance for studying the spatial imaging of nucleons and nuclei., Comment: 9 pages, 6 figures, 4 tables

Scientific software-defined as computer programs, scripts, or code used in scientific research, data analysis, modeling, or simulation-has become central to modern research. However, there is limited research on the readability and understandability of scientific code, both of which are vital for effective collaboration and reproducibility in scientific research. This study surveys 57 research scientists from various disciplines to explore their programming backgrounds, practices, and the challenges they face regarding code readability. Our findings reveal that most participants learn programming through self-study or on the-job training, with 57.9% lacking formal instruction in writing readable code. Scientists mainly use Python and R, relying on comments and documentation for readability. While most consider code readability essential for scientific reproducibility, they often face issues with inadequate documentation and poor naming conventions, with challenges including cryptic names and inconsistent conventions. Our findings also show low adoption of code quality tools and a trend towards utilizing large language models to improve code quality. These findings offer practical insights into enhancing coding practices and supporting sustainable development in scientific software., Comment: The 33rd IEEE/ACM International Conference on Program Comprehension (ICPC 2025) - Early Research Achievements Track

Assuming Galactic cosmic rays originate in supernovae and the winds of massive stars, starburst galaxies should produce very-high-energy (VHE; E$>$100 GeV) gamma-ray emission via the interaction of their copious quantities of cosmic rays with the large reservoirs of dense gas within the galaxies. Such VHE emission was detected by VERITAS from the starburst galaxy M 82 in 2008-09. An extensive, multi-year campaign followed these initial observations, yielding a total of 254 h of good quality VERITAS data on M 82. Leveraging modern analysis techniques and the larger exposure, these VERITAS data show a more statistically significant VHE signal ($\sim$6.5 standard deviations ($\sigma$)). The corresponding photon spectrum is well fit by a power law ($\Gamma = 2.3 \pm 0.3_{stat} \pm0.2_{sys}$) and the observed integral flux is F($>$450 GeV) = $(3.2 \pm0.6_{stat} \pm 0.6_{sys}) \times 10^{-13}~\mathrm{cm^{-2}~s}^{-1}$, or $\sim$0.4\% of the Crab Nebula flux above the same energy threshold. The improved VERITAS measurements, when combined with various multi-wavelength data, enable modeling of the underlying emission and transport processes. A purely leptonic scenario is found to be a poor representation of the gamma-ray spectral energy distribution (SED). A lepto-hadronic scenario with cosmic rays following a power-law spectrum in momentum (index $s\simeq 2.25$), and with significant bremsstrahlung below $1$~GeV, provides a good match to the observed SED. The synchrotron emission from the secondary electrons indicates that efficient non-radiative losses of cosmic-ray electrons may be related to advective escape from the starburst core., Comment: 15 pages, 7 figures; Accepted for the publication in The Astrophysical Journal (ApJ)

Blind face video restoration aims to restore high-fidelity details from videos subjected to complex and unknown degradations. This task poses a significant challenge of managing temporal heterogeneity while at the same time maintaining stable face attributes. In this paper, we introduce a Discrete Prior-based Temporal-Coherent content prediction transformer to address the challenge, and our model is referred to as DP-TempCoh. Specifically, we incorporate a spatial-temporal-aware content prediction module to synthesize high-quality content from discrete visual priors, conditioned on degraded video tokens. To further enhance the temporal coherence of the predicted content, a motion statistics modulation module is designed to adjust the content, based on discrete motion priors in terms of cross-frame mean and variance. As a result, the statistics of the predicted content can match with that of real videos over time. By performing extensive experiments, we verify the effectiveness of the design elements and demonstrate the superior performance of our DP-TempCoh in both synthetically and naturally degraded video restoration.

Multi-armed bandits (MAB) are commonly used in sequential online decision-making when the reward of each decision is an unknown random variable. In practice however, the typical goal of maximizing total reward may be less important than minimizing the total cost of the decisions taken, subject to a reward constraint. For example, we may seek to make decisions that have at least the reward of a reference ``default'' decision, with as low a cost as possible. This problem was recently introduced in the Multi-Armed Bandits with Cost Subsidy (MAB-CS) framework. MAB-CS is broadly applicable to problem domains where a primary metric (cost) is constrained by a secondary metric (reward), and the rewards are unknown. In our work, we address variants of MAB-CS including ones with reward constrained by the reward of a known reference arm or by the subsidized best reward. We introduce the Pairwise-Elimination (PE) algorithm for the known reference arm variant and generalize PE to PE-CS for the subsidized best reward variant. Our instance-dependent analysis of PE and PE-CS reveals that both algorithms have an order-wise logarithmic upper bound on Cost and Quality Regret, making our policies the first with such a guarantee. Moreover, by comparing our upper and lower bound results we establish that PE is order-optimal for all known reference arm problem instances. Finally, experiments are conducted using the MovieLens 25M and Goodreads datasets for both PE and PE-CS revealing the effectiveness of PE and the superior balance between performance and reliability offered by PE-CS compared to baselines from the literature.

Let $M$ be a Carath\'eodory hyperbolic complex manifold. We show that $M$ supports a real-analytic bounded strictly plurisubharmonic function. If $M$ is also complete K\"ahler, we show that $M$ admits the Bergman metric. When $M$ is strongly Carath\'eodory hyperbolic and is the universal covering of a quasi-projective manifold $X$, the Bergman metric can be estimated in terms of a Poincar\'e type metric on $X$. It is also proved that any quasi-projective (resp. projective) subvariety of $X$ is of log-general type (resp. general type), a result consistent with a conjecture of Lang., Comment: May be slightly different from published version

Traditional in-person psychological counseling remains primarily niche, often chosen by individuals with psychological issues, while online automated counseling offers a potential solution for those hesitant to seek help due to feelings of shame. Cognitive Behavioral Therapy (CBT) is an essential and widely used approach in psychological counseling. The advent of large language models (LLMs) and agent technology enables automatic CBT diagnosis and treatment. However, current LLM-based CBT systems use agents with a fixed structure, limiting their self-optimization capabilities, or providing hollow, unhelpful suggestions due to redundant response patterns. In this work, we utilize Quora-like and YiXinLi single-round consultation models to build a general agent framework that generates high-quality responses for single-turn psychological consultation scenarios. We use a bilingual dataset to evaluate the quality of single-response consultations generated by each framework. Then, we incorporate dynamic routing and supervisory mechanisms inspired by real psychological counseling to construct a CBT-oriented autonomous multi-agent framework, demonstrating its general applicability. Experimental results indicate that AutoCBT can provide higher-quality automated psychological counseling services.

The rapid evolution of communication technologies and the emergence of sixth-generation (6G) networks have introduced unprecedented opportunities for ultra-reliable, low-latency, and energy-efficient communication. However, the integration of advanced technologies like non-orthogonal multiple access (NOMA) and wireless powered communication networks (WPCNs) brings significant challenges, particularly in terms of energy constraints and security vulnerabilities. Traditional antenna systems and orthogonal multiple access schemes struggle to meet the increasing demands for performance and security in such environments. To address this gap, this paper investigates the impact of emerging fluid antenna systems (FAS) on the performance of physical layer security (PLS) in WPCNs. Specifically, we consider a scenario in which a transmitter, powered by a power beacon via an energy link, transmits confidential messages to legitimate FAS-aided users over information links while an external eavesdropper attempts to decode the transmitted signals. Additionally, users leverage the NOMA scheme, where the far user may also act as an internal eavesdropper. For the proposed model, we first derive the distributions of the equivalent channels at each node and subsequently obtain compact expressions for the secrecy outage probability (SOP) and average secrecy capacity (ASC), using the Gaussian quadrature methods. Our results reveal that incorporating the FAS for NOMA users, instead of the TAS, enhances the performance of the proposed secure WPCN.

As Large Language Models (LLMs) are integrated into electronic health record (EHR) workflows, validated instruments are essential to evaluate their performance before implementation. Existing instruments for provider documentation quality are often unsuitable for the complexities of LLM-generated text and lack validation on real-world data. The Provider Documentation Summarization Quality Instrument (PDSQI-9) was developed to evaluate LLM-generated clinical summaries. Multi-document summaries were generated from real-world EHR data across multiple specialties using several LLMs (GPT-4o, Mixtral 8x7b, and Llama 3-8b). Validation included Pearson correlation for substantive validity, factor analysis and Cronbach's alpha for structural validity, inter-rater reliability (ICC and Krippendorff's alpha) for generalizability, a semi-Delphi process for content validity, and comparisons of high-versus low-quality summaries for discriminant validity. Seven physician raters evaluated 779 summaries and answered 8,329 questions, achieving over 80% power for inter-rater reliability. The PDSQI-9 demonstrated strong internal consistency (Cronbach's alpha = 0.879; 95% CI: 0.867-0.891) and high inter-rater reliability (ICC = 0.867; 95% CI: 0.867-0.868), supporting structural validity and generalizability. Factor analysis identified a 4-factor model explaining 58% of the variance, representing organization, clarity, accuracy, and utility. Substantive validity was supported by correlations between note length and scores for Succinct (rho = -0.200, p = 0.029) and Organized ($\rho = -0.190$, $p = 0.037$). Discriminant validity distinguished high- from low-quality summaries ($p < 0.001$). The PDSQI-9 demonstrates robust construct validity, supporting its use in clinical practice to evaluate LLM-generated summaries and facilitate safer integration of LLMs into healthcare workflows.

Conflicting objectives present a considerable challenge in interleaving multi-task learning, necessitating the need for meticulous design and balance to ensure effective learning of a representative latent data space across all tasks without mutual negative impact. Drawing inspiration from the concept of marginal and conditional probability distributions in probability theory, we design a principled and well-founded approach to disentangle the original input into marginal and conditional probability distributions in the latent space of a variational autoencoder. Our proposed model, Deep Disentangled Interleaving Variational Encoding (DeepDIVE) learns disentangled features from the original input to form clusters in the embedding space and unifies these features via the cross-attention mechanism in the fusion stage. We theoretically prove that combining the objectives for reconstruction and forecasting fully captures the lower bound and mathematically derive a loss function for disentanglement using Na\"ive Bayes. Under the assumption that the prior is a mixture of log-concave distributions, we also establish that the Kullback-Leibler divergence between the prior and the posterior is upper bounded by a function minimized by the minimizer of the cross entropy loss, informing our adoption of radial basis functions (RBF) and cross entropy with interleaving training for DeepDIVE to provide a justified basis for convergence. Experiments on two public datasets show that DeepDIVE disentangles the original input and yields forecast accuracies better than the original VAE and comparable to existing state-of-the-art baselines.