Category
583 papers
Aesthetic preferences for AI-assisted smile designs differed meaningfully between software platforms and, to a lesser extent, between dental specialists and laypeople, with specialists showing a distinctly stronger relative preference for Smilecloud.
This small study demonstrated that DM-101PX was generally well tolerated and induced a strong and functional immune response.
Right ventricular free-wall longitudinal strain is a robust determinant of adverse outcomes beyond conventional LV indices and A4C-LVLS in VFMR and might be incorporated into routine evaluation to improve risk stratification in VFMR.
This record contains the replication package for the manuscript “Beyond Plans and Permissions: Enforcing Physical Effect Closure in Infrastructure Automation.” The package provides the EffectSeal reference implementation, executable verification tests, controlled Terraform/AWS-provider experiment configurations and frozen evidence, historical AWS Lambda and S3 provider reproductions, the AWS authorization-projection census and validity-hardening data, counterfactual stability and divergence analyses, approval-contract renderings, operational-cost accounting, and ordered-monitor scalability measurements. It also contains the frozen genuine-AWS CloudWatch Logs mediation witness and sanitized AWS CloudTrail corroboration. Reproduction instructions are provided in the package README and reproduction guide. The artifact supports the bounded empirical and reproducibility claims reported in the manuscript; it is not presented as production-wide cloud deployment validation or as a measurement of production end-to-end overhead.Licensing: Original EffectSeal software is released under the MIT License. Original data, documentation, and research artifacts produced by the authors are released under the Creative Commons Attribution 4.0 International License. Third-party materials and software remain subject to their respective copyright and license terms and are not relicensed by the authors.
Reach audiences
Advertise in front of researchers, engineers, and readers.
Supporting data for Mishra & Tyagi, "Analysis of Major Forbush Decrease Events during Current Solar Cycle in Association with Interplanetary and Geomagnetic Disturbances", submitted to Journal of Geophysical Research: Space Physics. This deposit contains the event catalogue, methodology documentation and validation dataset underlying the study. Three files are included. Supplementary_Data_S3_FD_Catalogue_265events.csv — A catalogue of 265 Forbush Decrease events spanning 4 December 2018 to 24 July 2026, detected automatically from hourly pressure-corrected count rates at the Oulu and Moscow neutron monitors using a rolling 48-hour baseline, and retained only where independently confirmed at both stations. Each event gives onset time, per-station minimum times and magnitudes, two-station mean magnitude, and severity class (211 weak, k=2; 43 moderate, k=3; 11 severe, k=4), together with matched interplanetary and geomagnetic parameters from OMNI: peak solar wind speed, peak total magnetic field, minimum Dst, maximum Kp, peak southward Bz, and elapsed time since the preceding disturbance. Matched OMNI parameters are available for 264 of the 265 events; one weak-class event falls within an OMNI data gap. A preceding disturbance is identified for 189 events. Supplementary_Data_S4_Methodology.md — Documentation sufficient to reproduce the analysis independently: input data sources, the event-detection algorithm and its thresholds, interplanetary parameter-matching windows, the severity-scoring framework, the statistical validation procedures, and the software used. Includes a file manifest and the disturbance-population test counts as computed. Supplementary_Data_S5_Disturbance_Episodes.csv — 369 interplanetary disturbance episodes detected independently of the FD catalogue, spanning 8 December 2018 to 3 July 2026, of which 218 produced no detectable Forbush Decrease. For each episode: onset and end times, duration, peak solar wind speed and magnetic field, minimum Dst, maximum Kp, whether an FD was produced and its matched severity and magnitude, the kinematic and magnetic component scores, the two-component severity score, and the resulting classification outcome. This file supports the false-alarm and recall analysis reported in the paper. All analysis used open-source Python libraries. Neutron monitor data are from the Oulu Cosmic Ray Station, IZMIRAN and the Neutron Monitor Database (www.nmdb.eu); interplanetary and geomagnetic parameters are from NASA OMNIWeb.
University technology campuses contain specialized laboratories, academic programs, and services that can be difficult for first-time visitors to identify. This paper presents INDI, a custom mobile campus guide robot that combines spoken interaction, synthesized speech, touchscreen feedback, animated facial states, head motion, and predefined mobile guidance behaviors. The platform retains the modular mechanical concept of an earlier prototype while replacing its Raspberry Pi and open-loop remote-control architecture with an NVIDIA Jetson Nano, an Arduino Uno motor-control bridge, ROS 1 nodes, encoder feedback, and dual PID speed loops. The robot weighs 2.37 kg, measures 39×27×69.5 cm, reaches a software-limited maximum speed of 0.4 m/s, and provides 27 min of continuous operation in the reported tests. Ten repetitions of each motion test produced mean displacements of 1.058 m and 2.182 m for 1 m and 2 m commands, respectively, and mean rotations of 89.2° and 180.3° for 90° and 180° commands. Voice trials achieved 90% correct interaction in a quiet environment and 70% under nearby conversational noise. In an exploratory user study with 13 participants, 16 of 20 assigned tasks were completed and the mean overall rating was 4.31/5. The results demonstrate the feasibility of an integrated, modular, physically embodied information service, while also revealing accumulated linear-motion error, sensitivity to ambient speech, limited battery duration, and the need for grounded institutional knowledge and autonomous localization. These findings are presented as preliminary evidence of technical and interaction feasibility rather than as confirmatory evidence of usability or campus-scale autonomous navigation.
The recent trend of Software-Defined Networking (SDN) has posed significant cybersecurity challenges as a result of its centralized control architecture, dynamic traffic behavior, and high programmability. Although these attributes improve network flexibility and management, they also increase vulnerability to Distributed Denial-of-Service (DDoS) attacks that can overwhelm network resources and disrupt services. Traditional signature- and rule-based detection methods may struggle with evolving traffic patterns and generate excessive false alarms. Machine learning offers a more promising solution that can learn the complex traffic patterns and separate malicious traffic from normal traffic. Most machine learning models, however, are black-box models that provide only superficial insight into the model predictions. Explainable Artificial Intelligence (XAI) addresses this limitation by identifying influential traffic features and providing interpretable evidence for detection decisions. This research develops an explainable machine learning-based framework for accurate, transparent, and reliable DDoS attack detection in an SDN environment. Several machine learning models are assessed, and XAI techniques are applied to explain the results of the predictions at global and instance levels. Gradient Boosting, Logistic Regression, AdaBoost, and Gaussian Naive Bayes were evaluated on 104,345 network-flow records using a 70:30 training–testing split. Gradient Boosting achieved the strongest performance, with 99.88% training accuracy, 99.87% testing accuracy, a testing F1-score of 99.84%, and a 0.20% miss rate. SHAP identified the most influential traffic features, while LIME linked individual predictions to feature-specific contributions. The proposed framework therefore combines reliable DDoS detection with transparent, analyst-oriented decision support for SDN security monitoring.
HEAR (Human-Evidence-driven Alignment of Requirements) is a requirements engineering framework for building inclusive digital health software. In HEAR, an evidence-attributed requirement leads every iteration: it carries the evidence that raised it, serves as the criterion the build is judged against, and receives the verdict of evaluation as a versioned change. The package holds 3 parts, two complete replication projects and a reuse kit. The first project is the paper's case study. A medication-management application for older adults in Australia taken through four HEAR iterations. It contains the backlog and its realignments at each of the four versions, the personas and cognitive walkthrough tasks, the workshop guideline and questionnaires, the requirement-driven test cases and evidence-driven defects for every version, and the evidence consumed in each iteration, including survey data, walkthrough data, and coded workshop data.The second project is the generalisation study. A redesign branch of the diet-tracking application MyFitnessPal reconstructed from a frozen 2020 baseline and executed largely by an LLM agent under human prioritisation gates, with the same artefacts across its three versions.The third folder holds the framework diagram and instructions for running HEAR on a new project. Workshop recordings and transcripts are withheld under ethics approval ERM49124. 1_case_study/ medication management for older adults, four iterations iteration_map.csv evidence, instrument, realignment, and resulting version per iteration 1_backlog/ the 33 requirements with their realignments, v1 to v4 2_personas_cw/ persona specifications and cognitive walkthrough tasks 3_workshop/ workshop guideline, task scripts, observation record, questionnaires 4_test_cases/ requirement-driven test cases, v1 to v4 5_evidence_bugs/ evidence-driven defects, v1 to v4 6_evidence_data/ survey data, walkthrough data, and coded workshop data per iteration 2_generalisation/ MyFitnessPal redesign branch, three iterations iteration_map.csv as above, including the frozen v0 baseline and each evidence window 1_backlog/ the redesign backlog with its realignments, v1 to v3 2_personas_cw/ persona specifications and the frozen walkthrough task set 3_test_cases/ requirement-driven test cases, v1 to v3 4_evidence_bugs/ evidence-driven defects, v1 to v3 5_evidence_data/ app review data, coding method, and coded results per iteration 3_use_hear/ for teams running HEAR on their own project HEAR_framework.drawio the full framework diagram how_to_use_HEAR.md step-by-step instructions
Semantic Contract Engineering (SCE) is an end-to-end engineering methodology for designing software systems through explicit semantic contracts established before implementation. SCE treats system meaning, intent, domain concepts, architectural structure, component responsibilities, interfaces, behavioral constraints, and technical specifications as explicit engineering artifacts. Implementation is consequently treated as a downstream realization of previously defined semantic and architectural decisions rather than as the primary place where system meaning is discovered. The methodology separates the development process into explicit stages while maintaining a layered software architecture in which components communicate through defined contracts. This separation is intended to improve traceability, consistency, testability, replaceability, and auditability across the system lifecycle. SCE is proposed as a methodological framework for engineering complex software systems, particularly systems in which human requirements, domain semantics, architectural decisions, computational models, and AI-assisted development must remain aligned and inspectable. This document presents the conceptual foundations, development stages, architectural principles, semantic contracts, and engineering model underlying Semantic Contract Engineering. .
Objective To understand the level of e-health literacy among sports majors and its influencing factors, providing reference for developing measures to improve e-health literacy in this population. Methods An online survey was conducted among college students from 6 universities in Jiangsu Province using an e-health literacy questionnaire. SPSS 26.0 statistical software was used for χ 2 test. t -test, one-way ANOVA, and multiple linear regression analysis were employed to explore the influencing factors of e-health literacy level. Results Among the 923 respondents, 722 (78.2%) were male and 201 (21.8%) were female. The average e-health literacy score was 41.91 ± 6.90, with an adequate rate of 22.2% (205 individuals). Influencing factor analysis showed that ethnicity (95%CI: −3.641 ~ −0.408), gender (95%CI: 1.292 ~ 3.445), self-rated health status compared with peers (95%CI: −1.276 ~ −0.100), use of smart wearable devices (95%CI: 0.100 ~ 0.794), harmony with surrounding people (95%CI: 0.379 ~ 1.733), frequency of searching health knowledge on the Internet (95%CI: 0.246 ~ 1.099), computer proficiency (95%CI: 0.230 ~ 1.569), days waking up feeling energetic in the morning (95%CI: 0.188 ~ 1.207), frequency of consuming fish, poultry, meat and eggs (95%CI: 0.194 ~ 1.331), and types of health knowledge content searched (95%CI: 0.023 ~ 1.556) were significant factors affecting e-health literacy among college students. Conclusion The e-health literacy level of sports majors is relatively low, with associated factors spanning multiple dimensions. Targeted interventions should prioritize male students, those with poorer self-rated health, and those with weaker interpersonal relationships. Strengthening digital skills and active information practice is a key strategy for improving e-health literacy in this population.
This repository contains the data products, statistical outputs, analysis scripts, plotting scripts, file-verification information, and software-environment records supporting the study “Hemispheric differences in eddy–jet relationships under Arctic and Antarctic sea-ice–SST perturbations.”The study uses large-ensemble PAMIP experiments from HadGEM3-GC31-MM and IPSL-CM6A-LR, with 100 members for each baseline and prescribed polar sea-ice–SST perturbation experiment. The main analyses compare Northern Hemisphere JJA and Southern Hemisphere DJF responses and diagnose approximately 2–8-day transient eddy kinetic energy (EKE), eddy momentum-flux convergence (EMFC), poleward eddy heat flux, jet latitude, storm-track–jet geometry, and the cross-experiment retention of baseline EMFC–jet relationships.The archive includes locked member inventories, SHA-256 verification information, polar boundary-state diagnostics, member-level derived metrics, statistical summary tables, reproducible analysis and figure-generation scripts, and software-environment information used for the final manuscript results. It is intended to support reproducibility of the reported ensemble-mean responses, hemispheric contrasts, sensitivity analyses, cross-experiment regression tests, and short-lag correspondence diagnostics.The original PAMIP model output is publicly available through the Earth System Grid Federation (ESGF) and is therefore not redistributed in this repository. Users should obtain the original HadGEM3-GC31-MM and IPSL-CM6A-LR PAMIP files directly from ESGF when full reconstruction from the raw model output is required.The repository should be cited together with the associated research article when the archived data products or code are reused.
This paper investigates the application of Multi-Agent Reinforcement Learning (MARL) to optimize software development processes. Traditional software development methodologies often struggle with adaptability and efficiency, particularly in complex projects. This research proposes a novel approach leveraging MARL to dynamically adjust and refine the development workflow. The core idea involves modeling the software development process as a multi-agent system, where each agent represents a distinct stage or activity. These agents learn optimal strategies through interaction and reward signals, leading to improved development efficiency and quality. We present a framework for formulating this problem, detailing the agent architecture, state space, action space, and reward function. The effectiveness of the MARL approach is demonstrated through a theoretical analysis and conceptual design, highlighting its potential to overcome limitations of static, rule-based methodologies. Future work will focus on implementing and testing this framework within a simulated software development environment.
In his latest book, Professor Eugene Fitzgerald examines the forces that turn breakthroughs into value — and why innovation resists simple formulas.
MIT students designed, built, and tested a jet engine with AI copilots, assessing AI’s usefulness in developing high-performance aerospace systems.
Researchers developed an auditing technique to test generative AI models for malicious capabilities, without prompting them for illegal outputs.
A USAF cadet and a Lincoln Laboratory researcher found AI chatbots can help nontechnical service members produce viable software applications for their unique problems.
Our latest voice model has improved precision and lower latency to make voice interactions more fluid, natural and precise.
Education Innovation
