Effective biomedical data integration depends on automated term normalization, the mapping of natural language biomedical terms to standardized identifiers. This linking of terms to identifiers is essential for semantic interoperability. Large language models (LLMs) show promise for this task but perform unevenly across terminologies. We evaluated both memorization (training-term performance) and generalization (validation-term performance) across multiple biomedical ontologies. Fine-tuning Llama 3.1 8B revealed marked differences by terminology. GO mappings showed strong memorization gains (up to 77% improvement in term-to-identifier accuracy), whereas HPO showed minimal improvement. Generalization occurred only for protein-gene (GENE) mappings (13.9% gain), while fine-tuning for HPO and GO yielded negligible transfer. Baseline accuracy varied by model scale, with GPT-4o outperforming both Llama variants for all terminologies. Embedding analyses showed tight semantic alignment between gene symbols and protein names but weak alignment between terms and identifiers for GO or HPO, consistent with limited lexicalization. Fine-tuning success depended on two interacting factors: identifier popularity and lexicalization. Popular identifiers were more likely encountered during pretraining, enhancing memorization. Lexicalized identifiers, such as gene symbols, enabled semantic generalization. By contrast, arbitrary identifiers in GO and HPO constrained models to rote learning. These findings provide a predictive framework for when fine-tuning enhances factual recall versus when it fails due to sparse or non-lexicalized identifiers.

We introduce a compact simulation framework for modeling open quantum systems coupled to structured, memory-retaining baths using QuTiP. Our method models the bath as a finite set of layered qubits with adjustable connections, interpreted either as a physical realization or as a conceptual representation, rather than as a continuum. This explicit modeling enables direct control over non-Markovian dynamics and allows spectral diagnostics via Fast Fourier Transform (FFT) of system observables. Using a triangle-based bath motif and its extension to a six-qubit anisotropic fractal-like architecture, we demonstrate how spectral fingerprints encode bath topology and memory depth. Standard machine learning tools such as Principal Component Analysis (PCA) and gradient boosting can then be employed to infer bath parameters and estimate proximity to exceptional points (EPs). The results suggest that spectral analysis can serve as a unifying, quantum-platform agnostic tool across theory, simulation, and experiment, offering both a student-accessible and experimentally relevant approach to understanding coherence loss and memory flow in quantum hardware. Rather than treating noise as an adversary to be eliminated, our approach views structured baths as collaborative partners, enabling controlled memory and delocalized memory and information flow for engineered quantum dynamics. In addition to its diagnostic power, the framework offers a modular and reproducible platform for teaching open quantum systems. Ultimately, we frame this as a pedagogical tool: students can pair FFT-based spectral features with lightweight ML (e.g., PCA and gradient boosting) to extract data-rich, interpretable signatures of open-system and non-Hermitian dynamics.

Distributed decision-making in multi-agent systems presents difficult challenges for interactive behavior learning in both cooperative and competitive systems. To mitigate this complexity, MAIDRL presents a semi-centralized Dense Reinforcement Learning algorithm enhanced by agent influence maps (AIMs), for learning effective multi-agent control on StarCraft Multi-Agent Challenge (SMAC) scenarios. In this paper, we extend the DenseNet in MAIDRL and introduce semi-centralized Multi-Agent Dense-CNN Reinforcement Learning, MAIDCRL, by incorporating convolutional layers into the deep model architecture, and evaluate the performance on both homogeneous and heterogeneous scenarios. The results show that the CNN-enabled MAIDCRL significantly improved the learning performance and achieved a faster learning rate compared to the existing MAIDRL, especially on more complicated heterogeneous SMAC scenarios. We further investigate the stability and robustness of our model. The statistics reflect that our model not only achieves higher winning rate in all the given scenarios but also boosts the agent's learning process in fine-grained decision-making.

We establish a deterministic and stochastic spherical quasi-interpolation framework featuring scaled zonal kernels derived from radial basis functions on the ambient Euclidean space. The method incorporates both quasi-Monte Carlo and Monte Carlo quadrature rules to construct easily computable quasi-interpolants, which provide efficient approximation to Sobolev-space functions for both clean and noisy data. To enhance the approximation power and robustness of our quasi-interpolants, we develop a multilevel method in which quasi-interpolants constructed with graded resolutions join force to reduce the error of approximation. In addition, we derive probabilistic concentration inequalities for our quasi-interpolants in pertinent stochastic settings. The construction of our quasi-interpolants does not require solving any linear system of equations. Numerical experiments show that our quasi-interpolation algorithm is more stable and robust against noise than comparable ones in the literature.

Centralized electronic health record repositories are critical for advancing disease surveillance, public health research, and evidence-based policymaking. However, developing countries face persistent challenges in achieving this due to fragmented healthcare data sources, inconsistent record-keeping practices, and the absence of standardized patient identifiers, limiting reliable record linkage, compromise data interoperability, and limit scalability-obstacles exacerbated by infrastructural constraints and privacy concerns. To address these barriers, this study proposes a scalable, privacy-preserving clinical data warehouse, NCDW, designed for heterogeneous EHR integration in resource-limited settings and tested with 1.16 million clinical records. The framework incorporates a wrapper-based data acquisition layer for secure, automated ingestion of multisource health data and introduces a soundex algorithm to resolve patient identity mismatches in the absence of unique IDs. A modular data mart is designed for disease-specific analytics, demonstrated through a dengue fever case study in Bangladesh, integrating clinical, demographic, and environmental data for outbreak prediction and resource planning. Quantitative assessment of the data mart underscores its utility in strengthening national decision-support systems, highlighting the model's adaptability for infectious disease management. Comparative evaluation of database technologies reveals NoSQL outperforms relational SQL by 40-69% in complex query processing, while system load estimates validate the architecture's capacity to manage 19 million daily records (34TB over 5 years). The framework can be adapted to various healthcare settings across developing nations by modifying the ingestion layer to accommodate standards like ICD-11 and HL7 FHIR, facilitating interoperability for managing infectious diseases (i.e., COVID, tuberculosis).

For radial basis function (RBF) kernel interpolation of scattered data, Schaback in 1995 proved that the attainable approximation error and the condition number of the underlying interpolation matrix cannot be made small simultaneously. He referred to this finding as an "uncertainty relation", an undesirable consequence of which is that RBF kernel interpolation is susceptible to noisy data. In this paper, we propose and study a distributed interpolation method to manage and quantify the uncertainty brought on by interpolating noisy spherical data of non-negligible magnitude. We also present numerical simulation results showing that our method is practical and robust in terms of handling noisy data from challenging computing environments.

TIC033834484 and TIC309658435 are long-period pulsating subdwarf B star, which were observed extensively (675 and 621 days, respectively) by the Transiting Exoplanet Survey Satellite (TESS). The high-precision photometric light curve reveals the presence of more than 40 pulsation modes including both stars. All the oscillation frequencies that we found are associated with gravity (g)-mode pulsations, with frequencies spanning from 80 $\mu$Hz (2 500 s) to 400 $\mu$Hz (12 000 s). We utilize the asteroseismic tools including asymptotic period spacings and rotational frequency multiplets in order to identify the pulsational modes. We found dipole (l = 1) mode sequences for both targets and calculate the mean period spacing of dipole modes ($\Delta P_{l=1}$), which allows us to identify the modes. Frequency multiplets provide a rotation period of about 64 d for TIC033834484. From follow-up ground-based spectroscopy, we find that TIC\,033834484 has an effective temperature of 24 210 K (140), a surface gravity of logg = 5.28 (03) and TIC309658435 has an effective temperature of 25 910 K (150), a surface gravity of logg = 5.48 (03).

Quality of Life (QOL) outcomes are important in the management of chronic illnesses. In studies of efficacies of treatments or intervention modalities, QOL scales multi-dimensional constructs are routinely used as primary endpoints. The standard data analysis strategy computes composite (average) overall and domain scores, and conducts a mixed-model analysis for evaluating efficacy or monitoring medical conditions as if these scores were in continuous metric scale. However, assumptions of parametric models like continuity and homoscedastivity can be violated in many cases. Furthermore, it is even more challenging when there are missing values on some of the variables. In this paper, we propose a purely nonparametric approach in the sense that meaningful and, yet, nonparametric effect size measures are developed. We propose estimator for the effect size and develop the asymptotic properties. Our methods are shown to be particularly effective in the presence of some form of clustering and/or missing values. Inferential procedures are derived from the asymptotic theory. The Asthma Randomized Trial of Indoor Wood Smoke data will be used to illustrate the applications of the proposed methods. The data was collected from a three-arm randomized trial which evaluated interventions targeting biomass smoke particulate matter from older model residential wood stoves in homes that have children with asthma.

This paper provides a review of recent publications and working papers on ChatGPT and related Large Language Models (LLMs) in accounting and finance. The aim is to understand the current state of research in these two areas and identify potential research opportunities for future inquiry. We identify three common themes from these earlier studies. The first theme focuses on applications of ChatGPT and LLMs in various fields of accounting and finance. The second theme utilizes ChatGPT and LLMs as a new research tool by leveraging their capabilities such as classification, summarization, and text generation. The third theme investigates implications of LLM adoption for accounting and finance professionals, as well as for various organizations and sectors. While these earlier studies provide valuable insights, they leave many important questions unanswered or partially addressed. We propose venues for further exploration and provide technical guidance for researchers seeking to employ ChatGPT and related LLMs as a tool for their research.

Suicide is a critical global health problem involving more than 700,000 deaths yearly, particularly among young adults. Many people express their suicidal thoughts on social media platforms such as Reddit. This paper evaluates the effectiveness of the deep learning transformer-based models BERT, RoBERTa, DistilBERT, ALBERT, and ELECTRA and various Long Short-Term Memory (LSTM) based models in detecting suicidal ideation from user posts on Reddit. Toward this objective, we curated an extensive dataset from diverse subreddits and conducted linguistic, topic modeling, and statistical analyses to ensure data quality. Our results indicate that each model could reach high accuracy and F1 scores, but among them, RoBERTa emerged as the most effective model with an accuracy of 93.22% and F1 score of 93.14%. An LSTM model that uses attention and BERT embeddings performed as the second best, with an accuracy of 92.65% and an F1 score of 92.69%. Our findings show that transformer-based models have the potential to improve suicide ideation detection, thereby providing a path to develop robust mental health monitoring tools from social media. This research, therefore, underlines the undeniable prospect of advanced techniques in Natural Language Processing (NLP) while improving suicide prevention efforts.

Macroscopic quantum amplifiers maintain coherence even while strongly coupled to their surroundings, demonstrating that coherence can be preserved through architecture rather than isolation. Here we derive a finite structured-bath Hamiltonian in which dissipation and feedback originate from the same microscopic couplings. The resulting self-energy {\Sigma}({\omega}) exhibits coupled real and imaginary parts whose evolution reproduces the breathing dynamics observed in Josephson quantum amplifiers. This establishes quantum reciprocity: macroscopic coherence lives not in isolation, but in structured connection. We numerically validate this principle by engineering a six-qubit structured bath to demonstrate controllable transitions from dissipation to amplification. This architectural core serves as the foundation for a proposed multi-scale workflow to transform quantum noise into a design resource, preserving coherence not through isolation but through architectural reciprocity.

[Abridged] The Study Analysis Group 8 of the NASA Exoplanet Analysis Group was convened to assess the current capabilities and the future potential of the precise radial velocity (PRV) method to advance the NASA goal to "search for planetary bodies and Earth-like planets in orbit around other stars.: (U.S. National Space Policy, June 28, 2010). PRVs complement other exoplanet detection methods, for example offering a direct path to obtaining the bulk density and thus the structure and composition of transiting exoplanets. Our analysis builds upon previous community input, including the ExoPlanet Community Report chapter on radial velocities in 2008, the 2010 Decadal Survey of Astronomy, the Penn State Precise Radial Velocities Workshop response to the Decadal Survey in 2010, and the NSF Portfolio Review in 2012. The radial-velocity detection of exoplanets is strongly endorsed by both the Astro 2010 Decadal Survey "New Worlds, New Horizons" and the NSF Portfolio Review, and the community has recommended robust investment in PRVs. The demands on telescope time for the above mission support, especially for systems of small planets, will exceed the number of nights available using instruments now in operation by a factor of at least several for TESS alone. Pushing down towards true Earth twins will require more photons (i.e. larger telescopes), more stable spectrographs than are currently available, better calibration, and better correction for stellar jitter. We outline four hypothetical situations for PRV work necessary to meet NASA mission exoplanet science objectives.

We present an IR-monitoring survey with the $Spitzer$ Space Telescope of the star forming region GGD 12-15. Over 1000 objects were monitored including about 350 objects within the central 5 arcminutes which is found to be especially dense in cluster members. The monitoring took place over 38 days and is part of the Young Stellar Object VARiability (YSOVAR) project. The region was also the subject of a contemporaneous 67ks $Chandra$ observation. The field includes 119 previously identified pre-main sequence star candidates. X-rays are detected from 164 objects, 90 of which are identified with cluster members. Overall, we find that about half the objects in the central 5 arcminutes are young stellar objects based on a combination of their spectral energy distribution, IR variability and X-ray emission. Most of the stars with IR excess relative to a photosphere show large amplitude (>0.1 mag) mid-IR variability. There are 39 periodic sources, all but one of these is found to be a cluster member. Almost half of the periodic sources do not show IR excesses. Overall, more than 85% of the Class I, flat spectrum, and Class II sources are found to vary. The amplitude of the variability is larger in more embedded young stellar objects. Most of the Class~I/II objects exhibit redder colors in a fainter state, compatible with time-variable extinction. A few become bluer when fainter, which can be explained with significant changes in the structure of the inner disk. A search for changes in the IR due to X-ray events is carried out, but the low number of flares prevented an analysis of the direct impact of X-ray flares on the IR lightcurves. However, we find that X-ray detected Class II sources have longer timescales for change in the mid-IR than a similar set of non-X-ray detected Class IIs.

Neuromorphic computing, inspired by the human brain's neural architecture, is revolutionizing artificial intelligence and edge computing with its low-power, adaptive, and event-driven designs. However, these unique characteristics introduce novel cybersecurity risks. This paper proposes Neuromorphic Mimicry Attacks (NMAs), a groundbreaking class of threats that exploit the probabilistic and non-deterministic nature of neuromorphic chips to execute covert intrusions. By mimicking legitimate neural activity through techniques such as synaptic weight tampering and sensory input poisoning, NMAs evade traditional intrusion detection systems, posing risks to applications such as autonomous vehicles, smart medical implants, and IoT networks. This research develops a theoretical framework for NMAs, evaluates their impact using a simulated neuromorphic chip dataset, and proposes countermeasures, including neural-specific anomaly detection and secure synaptic learning protocols. The findings underscore the critical need for tailored cybersecurity measures to protect brain-inspired computing, offering a pioneering exploration of this emerging threat landscape.

Exoplanet systems with multiple planets in mean motion resonances have often been hailed as a signpost of disk driven migration. Resonant chains like Kepler-223 and Kepler-80 consist of a trio of planets with the three-body resonant angle librating and/or with a two-body resonant angle librating for each pair. Here we investigate whether close-in super-Earths and mini-Neptunes forming in situ can lock into resonant chains due to dissipation from a depleted gas disk. We simulate the giant impact phase of planet formation, including eccentricity damping from a gaseous disk, followed by subsequent dynamical evolution over tens of millions of years. In a fraction of simulated systems, we find that planets naturally lock into resonant chains. These planets achieve a chain of near-integer period ratios during the gas disk stage, experience eccentricity damping that captures them into resonance, stay in resonance as the gas disk dissipates, and avoid subsequent giant impacts, eccentricity excitation, and chaotic diffusion that would dislodge the planets from resonance. Disk conditions that enable planets to complete their formation during the gas disk stage enable those planets to achieve tight period ratios <= 2 and, if they happen to be near integer period ratios, lock into resonance. Using the weighting of different disk conditions deduced by MacDonald et al. (2020) and forward modeling Kepler selection effects, we find that our simulations of in situ formation via oligarchic growth lead to a rate of observable trios with integer period ratios and librating resonant angles comparable to observed Kepler systems.

Cloud computing has grown rapidly in recent years, mainly due to the sharp increase in data transferred over the internet. This growth makes load balancing a key part of cloud systems, as it helps distribute user requests across servers to maintain performance, prevent overload, and ensure a smooth user experience. Despite its importance, managing server resources and keeping workloads balanced over time remains a major challenge in cloud environments. This paper introduces a novel Score-Based Dynamic Load Balancer (SBDLB) that allocates workloads to virtual machines based on real-time performance metrics. The objective is to enhance resource utilization and overall system efficiency. The method was thoroughly tested using the CloudSim 7G platform, comparing its performance against the throttled load balancing strategy. Evaluations were conducted across a variety of workloads and scenarios, demonstrating the SBDLB's ability to adapt dynamically to workload fluctuations while optimizing resource usage. The proposed method outperformed the throttled strategy, improving average response times by 34% and 37% in different scenarios. It also reduced data center processing times by an average of 13%. Over a 24-hour simulation, the method decreased operational costs by 15%, promoting a more energy-efficient and sustainable cloud infrastructure through reduced energy consumption.

Data sites selected from modeling high-dimensional problems often appear scattered in non-paternalistic ways. Except for sporadic clustering at some spots, they become relatively far apart as the dimension of the ambient space grows. These features defy any theoretical treatment that requires local or global quasi-uniformity of distribution of data sites. Incorporating a recently-developed application of integral operator theory in machine learning, we propose and study in the current article a new framework to analyze kernel interpolation of high dimensional data, which features bounding stochastic approximation error by the spectrum of the underlying kernel matrix. Both theoretical analysis and numerical simulations show that spectra of kernel matrices are reliable and stable barometers for gauging the performance of kernel-interpolation methods for high dimensional data.

Multi-Agent Reinforcement Learning (MARL) algorithms are widely adopted in tackling complex tasks that require collaboration and competition among agents in dynamic Multi-Agent Systems (MAS). However, learning such tasks from scratch is arduous and may not always be feasible, particularly for MASs with a large number of interactive agents due to the extensive sample complexity. Therefore, reusing knowledge gained from past experiences or other agents could efficiently accelerate the learning process and upscale MARL algorithms. In this study, we introduce a novel framework that enables transfer learning for MARL through unifying various state spaces into fixed-size inputs that allow one unified deep-learning policy viable in different scenarios within a MAS. We evaluated our approach in a range of scenarios within the StarCraft Multi-Agent Challenge (SMAC) environment, and the findings show significant enhancements in multi-agent learning performance using maneuvering skills learned from other scenarios compared to agents learning from scratch. Furthermore, we adopted Curriculum Transfer Learning (CTL), enabling our deep learning policy to progressively acquire knowledge and skills across pre-designed homogeneous learning scenarios organized by difficulty levels. This process promotes inter- and intra-agent knowledge transfer, leading to high multi-agent learning performance in more complicated heterogeneous scenarios.

Recently, much attention has been devoted to finding highly efficient and powerful activation functions for CNN layers. Because activation functions inject different nonlinearities between layers that affect performance, varying them is one method for building robust ensembles of CNNs. The objective of this study is to examine the performance of CNN ensembles made with different activation functions, including six new ones presented here: 2D Mexican ReLU, TanELU, MeLU+GaLU, Symmetric MeLU, Symmetric GaLU, and Flexible MeLU. The highest performing ensemble was built with CNNs having different activation layers that randomly replaced the standard ReLU. A comprehensive evaluation of the proposed approach was conducted across fifteen biomedical data sets representing various classification tasks. The proposed method was tested on two basic CNN architectures: Vgg16 and ResNet50. Results demonstrate the superiority in performance of this approach. The MATLAB source code for this study will be available at this https URL