67471 research outputs found
Sort by
NEO-Grid: A Neural Approximation Framework for Optimization and Control in Distribution Grids
Get PDFThe rise of distributed energy resources (DERs) is reshaping modern distribution grids, introducing new challenges for maintaining voltage stability under dynamic and decentralized operating conditions. This paper presents NEO-Grid, a unified learning-based framework for volt-var optimization (VVO) and volt-var control (VVC) that leverages neural network surrogates for power flow and deep equilibrium models (DEQs) for closed-loop control. Our method replaces traditional linear approximations with piecewise-linear ReLU networks trained to capture the nonlinear relationship between power injections and voltage magnitudes. For control, we model the recursive interaction between voltage and inverter response using DEQs, allowing direct fixed-point computation and efficient training via implicit differentiation. We evaluate NEO-Grid on the IEEE 33-bus system, demonstrating that it significantly improves voltage regulation performance compared to standard linear and heuristic baselines in both optimization and control settings. Our results establish NEO-Grid as a scalable, accurate, and interpretable solution for learning-based voltage regulation in distribution grids
Investigating the Pareto Principle in Student Software Engineering Team Projects
Get PDFThe Pareto principle, or 80/20 rule, has been observed in open-source and professional software projects, indicating that ”vital few” 20% of contributors account for at least 80% of the outcomes. However, its applicability to academic software engineering teams remains uncertain. This study investigates contribution inequality among student developers in software engineering teams by analyzing data-mined GitHub activities. Using the Hoover Index and Lorenz Curve, we analyzed whether a small subset of students performed a disproportionate share of the work to their teams. Contrary to the classic Pareto distribution, our results show that out of 94 teams, only six teams (6.4%) met the 80/20 criterion—–where their top 20% of contributors produced at least 80% of that team’s work–—while the remaining 88 teams did not reach this threshold. We discuss potential academic factors such as grading incentives, peer review, and structured team dynamics that may mitigate contribution imbalance
Guess, Learn, Repeat: Intelligent Learning System with Synthetic and Counterfactual Training in a GeoGuessr-Inspired Classification Task
No full textTraining novices by experts is often costly and time-consuming. Alternatively, learning systems offer a scalable and automated alternative. However, learning systems offer another, yet underexplored advantage, over training with experts: Analyzing novices and providing personalized training. This study explores the use of synthetically generated images to improve novice image classification skills in a GeoGuessr-inspired classification task. By leveraging a counterfactual-based approach and synthetically generated personalized training data, we aim to enhance individual learning. In a controlled experiment where participants classify Google Street View images from four different cities, we compare the impact of personalized synthetic images against randomly assigned ones. Our findings indicate that personalized training improves classification accuracy, underscoring the potential of intelligent learning. These results highlight a promising direction for integrating synthetic data into adaptive training environments in game-like settings, paving the way for effective and personalized intelligent learning systems
Leadership That Clicks: The Impact of CIO Presence on Employees' Digital Performance in Organizations
No full textThis study explores how the presence of a Chief Information Officer (CIO) in an organization impacts employee digital performance. While prior research has focused on firm-level outcomes, this paper focuses on the employee level and investigates how CIOs influence both routine and innovative digital tasks of employees. The study proposes that CIO presence enhances IT knowledge sharing and improves employee digital performance within the organization. It further examines how a CEO’s IT background and the board’s R&D experience moderate CIOs’ influence on employees’ performance. The research employs surveys and firm-level data to measure CIO presence, IT knowledge sharing, and employee digital task and innovative performance using data from a broad range of U.S. firms across various sizes and industries. This study contributes to understanding the strategic value of CIOs in enabling digital transformation at the individual employee level
Understanding Physicians’ Continued Use of Robotic Surgical Navigation Systems: An Integrated TAM–TPB Perspective
Get PDFRobotic surgical navigation systems have become increasingly important in modern surgery, offering precision and consistency that enhance clinical outcomes. Despite their growing adoption, limited attention has been paid to the perspectives of physicians who directly operate these systems in clinical practice. To address this gap, this study investigates the determinants of physicians’ continued use of robotic surgical navigation systems by integrating the Technology Acceptance Model (TAM) and the Theory of Planned Behavior (TPB). Data were collected from 120 orthopedic surgeons across seven major medical centers in Taiwan, with 115 valid responses (96% response rate). Using Partial Least Squares Structural Equation Modeling (PLS-SEM), the results reveal that task complexity and social influence significantly affect perceived usefulness, while system self-efficacy and facilitating conditions influence perceived ease of use. Perceived ease of use indirectly impacts continuance intention through perceived usefulness, and top management support is identified as a critical reinforcing factor. These findings extend TAM and TPB to advanced surgical technologies and offer practical implications for promoting long-term adoption in clinical settings
Governing Manipulative and Synthetic Content on Social Media Platforms
Get PDFSocial media platforms are immensely popular among young adults (13-20 years old). However, these platforms also pose distinct risks, including fraudulent advertisements and scams. The rise of Generative AI (GenAI) based advertising has exacerbated these risks, including deepfake scams. Using the integrated Governing Knowledge Commons-Contextual Integrity (GKC-CI) framework, we perform a structured content analysis of the policies from three stakeholders: social media platforms, foundational GenAI model providers, and GenAI-based advertising services. We analyze how these entities define and constrain the use of GenAI in social media advertising, with particular attention to protection for young adults against manipulative and synthetic GenAI content. We found that current governance lacks enforceable rules regarding GenAI, and are especially lacking in protecting young adults. We discuss implications for platforms and regulators to strengthen institutional governance for GenAI-based advertising targeting young adults
Intelligent Agile: Conceptual Modeling and AI for Next-Generation Agile Software Development
Get PDFAlthough Agile is now the most popular software development methodology, agile projects continue to fail. Conceptual modeling can help address at least two major challenges in agile projects: managing requirements and communication for collaboration and coordination. At the same time, conceptual modeling has not been well received by agile teams, as traditional conceptual modeling comes with high overhead. We propose a new approach to agile that leverages artificial intelligence (AI) and conceptual modeling together. This design approach enables us to advance beyond passive conceptual modeling representations, such as static diagrams, to the new types of information systems (IS), intelligent conceptual modeling systems (ICMS), that have awareness, autonomy, adaptivity, and activity. In this way, conceptual modeling, powered by AI, facilitates requirements and communication in agile projects, resulting in what we call intelligent agile
Can AI Recognize Its Own Reflection? Self-Detection Performance of LLMs in Computing Education
Get PDFThe rapid advancement of Large Language Models (LLMs) presents a significant challenge to academic integrity within computing education. As educators seek reliable detection methods, this paper evaluates the capacity of three prominent LLMs (GPT-4, Claude, and Gemini) to identify AI-generated text in computing-specific contexts. We test their performance under both standard and `deceptive' prompt conditions, where the models were instructed to evade detection. Our findings reveal a significant instability: while default AI-generated text was easily identified, all models struggled to correctly classify human-written work (with error rates up to 32%). Furthermore, the models were highly susceptible to deceptive prompts, with Gemini's output completely fooling GPT-4. Given that simple prompt alterations significantly degrade detection efficacy, our results demonstrate that these LLMs are currently too unreliable for making high-stakes academic misconduct judgments
