February 27 ~ 28, 2026, Vancouver, Canada
Grace Llego1 Jim Alves-Foss2 1,2 Center for Secure and Dependable Systems, University of Idaho. Moscow, ID USA,
Since about 2010, there has been increased attention in gauging and improving digital citizenship across the world. One important aspect of digital citizenship is cybersecurity awareness. This paper reports on a review of 35 studies that have been conducted in this area, primarily consisting of surveys of students. The primary purpose of this review was to compare and evaluate the methods used in these studies to provide guidance towards future studies. The studies surveyed from 32 to over 3000 participants evaluated cybersecurity awareness and/or practice, with about half of them providing copies of the survey questions to allow for follow-on comparative studies.
Cyber security awareness, Surveys. Digital Literacy
Jo˜ao Vicente Markovicz Martins, Jonathan Santos da Silva, Rodrigo Ribeiro Barbosa da Silva, Giovanna Vict´oria Souza Venier, and Paulo Jorge Matos1 Polytechnic Institute of Bragan¸ca (IPB) School of Technology and Management, Bragan¸ca, PT
The landscape of data analytics is shifting towards high-performance local processing, necessitating efficient tools for handling substantial datasets. This paper evaluates the performance of DuckDB,an embedded Online Analytical Processing (OLAP) database, specifically focusing on its interaction with different file formats. We conduct a comparative analysis of data loading times, query execution speeds, and sorting efficiency across three widely used formats: Apache Parquet, CSV, and JSON. While row-oriented formats like CSV and JSON are ubiquitous for data interchange, our benchmarks demonstrate that they impose significant I/O overhead in analytical pipelines. Conversely, the results highlight DuckDB’s optimized handling of columnar formats, showing that Parquet offers superior performance in both ingestion and query latency. These findings reinforce the importance of format selection in local analytical workflows and validate DuckDB’s suitability for modern data engineering tasks..
DuckDB, data processing, Parquet, CSV, JSON, performance benchmarking, file formats
Huaqiu Wu 1, Tyler Boulom21 Webb Schools, 1175 W Baseline Rd, Claremont, CA 91711 2 Woodbury University, 7500 N Glenoaks Blvd, Burbank, CA 91504
This paper addresses the need for healthier and more meaningful digital experiences for children and seniors. Many modern games contribute to anxiety, overstimulation, and cognitive decline, while seniors face increased dementia risk from inactivity. Adventures of Weiqi offers a solution by transforming the ancient strategy game Go into an accessible, educational adventure. The system features three interconnected components: a training hub, an adventure progression system, and a review center. Development involved overcoming challenges in UI design, level organization, and AI implementation. Two experiments demonstrated that structured training improves puzzle accuracy and that reviewing past games enhances future performance. Overall, the project shows strong potential as a cognitive-training tool that blends entertainment with mental stimulation.
Cognitive Training, Game-Based Learning, Strategic Games, Digital Wellbeing
Rogan Songqi Wu1 Tyler Boulom2 , 1 1St Margaret’s Episcopal School, 31641 La Novia Ave, San Juan Capistrano, CA 92675 , 2 2Woodbury University, 7500 N Glenoaks Blvd, Burbank, CA 91504
Out-of-hospital cardiac arrest is a leading cause of preventable death, with survival decreasing by approximately 10% for every minute without cardiopulmonary resuscitation (CPR) [1]. Despite this, many bystanders hesitate to intervene due to limited access to realistic, hands-on training and low confidence in their skills. This paper introduces Vitals CPR, a virtual reality (VR)–based training game designed to provide immersive, repeatable CPR practice in a safe environment. Developed using Unity and SteamVR, the system places users in first-person emergency scenarios where they perform CPR with real-time visual, audio, and haptic feedback aligned with American Heart Association guidelines. Experimental testing evaluated pulse-check reliability and compression timing under natural hand movement. Results showed that targeted feedback adjustments improved accuracy, reduced user error, and increased confidence. Compared to traditional CPR instruction, Vitals CPR offers greater accessibility, engagement, and skill reinforcement [2]. While currently limited in scope, this project demonstrates the potential of VR as an effective tool for improving CPR readiness and empowering more people to act in life-threatening emergencies.
Virtual Reality Training, CPR Education, Emergency Response, Immersive Simulation
Jiawei Zhang2 Xin Zhang3 Xinyin Mia1 Senior Investment Analyst, PRA Group (Nasdaq: PRAA), Norfolk, Virginia, USA 2 Data Scientist, PRA Group (Nasdaq: PRAA), Norfolk, Virginia, USA3 Senior Data Analyst, American Airlines Group Inc (Nasdaq: AAL), Dallas, Texas,USA
This paper provides an innovative methodology of partial penalty on machine learning models to handle the data imbalance scenario occurring in credit card fraud detection implementation. Unlike the normal over-sampling or under-sampling methodologies, partial penalty directs the machine learning model to focus on learning the minor class of target variable even when the class distribution is extremely imbalanced. Besides comparing the partial penalty approach with over-sampling and under-sampling approaches to handle data imbalance scenario, we’ve implemented this new approach under five machine learning classification models, including Logistic Regression, Random Forest, kNN, Decision Tree, and Light Gradient Boosting Model. The new partial penalty approach realizes a performance of 88.35% F1 score and 98.79% AUC score with Light GBM, higher than either over-sampling or under-sampling approaches.
Partial Penalty, Gradient Boosting, Data Imbalance, Credit Card Fraud Detection, SMOTE
Jiabao Hu1 Ang Li2 , 1 SNortheastern University, 5000 MacArthur Boulevard, Oakland, CA 94613 2 California State University Long Beach, 1250 Bellflower Blvd, Long Beach, CA 90840
Small businesses face significant challenges by accessing professional expertise in marketing, customer service, financial planning, and strategic development. While large language models offer unprecedented capabilities for intelligent assistance, accessibility gaps prevent many entrepreneurs from benefiting. EntrepreneurHub addresses this problem through a cross-platform mobile application integrating OpenAI’s GPT model with Firebase backend services. The application implements domain-specific AI modules using specialized prompt engineering to ensure contextually appropriate responses. Key components include secure authentication, a centralized AI service layer, and usage analytics tracking. Experiments evaluated response quality across domains, achieving average scores of 4.54/5.0, and measured latency performance across network conditions. Comparison with existing methodologies demonstrates improvements in accessibility, cost, and functional breadth. Results validate that sophisticated AI capabilities can be effectively delivered through intuitive mobile interfaces, democratizing access to intelligent business guidance for entrepreneurs lacking technical expertise or significant budgets.
Large Language Models, Mobile Application Development, Flutter Framework, Small Business Technology, Artificial Intelligence
Chris Zheng1 Ang Li2 , 1 The Bishop’s School, 7607 La Jolla Blvd., La Jolla, CA 92037 2 California State Polytechnic University, Pomona, 3801 W Temple Ave, Pomona, CA 91768
In computer vision applications, the You Only Look Once (YOLO) model serves as an effective single shot detector for identifying and localizing objects of interest. However, in complex environments where objects are partially occluded or exhibit low contrast relative to their surroundings, YOLO and similar object detectors encounter significant challenges in accuracy and efficiency [3]. This study presents an enhanced model designed to improve the accuracy and efficiency of human detection in disaster aftermath scenarios, including earthquakes, wildfires, and collapsed buildings. The proposed approach integrates Spatial Transformer Networks (STNs) with YOLOv8 to enhance spatial invariance capabilities. A dataset comprising over 8,500 controlled images sourced from Roboflow Universe was utilized to train both a baseline YOLOv8 model and the experimental STN+YOLOv8 model. Through an iterative design process involving eight model variations, the results demonstrate that the STN+YOLOv8 configuration achieves a mean Average Precision (mAP) of 79.9% compared to 61.2% for the baseline model [4]. Additionally, an S500 drone platform equipped with customized 3Dprinted components was developed to demonstrate real time deployment feasibility for emergency response applications. Future work includes field deployment and integration of geolocation features for autonomous victim localization.
Object Detection, YOLOv8, Spatial Transformer Networks, Unmanned Aerial Vehicles, Search and Rescue, Computer Vision, Deep Learning, Disaster Response
Ashwin Geeni 1, Independent Researcher, USA
AI and deep learning workloads rely heavily on matrix multiplication, which demands both high speed and numerical accuracy. Traditional processors use fixed arithmetic precision, forcing a global trade-off between performance and accuracy. This paper presents the Mixed Precision Pipeline (MPP), a five-stage RISC-V processor pipeline that dynamically switches between INT8, FP16, and FP32 precisions based on runtime numerical stability analysis..
Mixed-precision computing, RISC-V, Matrix multiplication, Hardware acceleration, Numerical stability
Abbas Rashid Butt1, Laiba Abbas Butt2, Rizwan Ahmad3, Nameem Ullah Tariq4 and Azhar ul Haq Wahid5 , 1Ph.D. Scholar, University of Management and Science, Punjab, Pakistan 2BS Criminology and Security Studies, ISCS, Punjab University, Pakistan. 3 Media Consultant, FM 101 Radio Pakistan 4 PhD Scholar, School of Media and Communication SMC Shanghai Jiao Tong University, Shanghai, China. 55Universidad Complutense de Madrid (UCM), Spain.
Artificial intelligence has become a central force in global digital communications, shaping public discourse through algorithmic mediation and hybrid influence practices. Using a sequential explanatory mixed-methods design, this study examines how AI-mediated systems structure knowledge, shape perception, and drive strategic content dissemination. Qualitative analysis reveals that expert discourse is dominated by frequent concerns related to ethics, law, privacy and bias, as well as references to AI and algorithms, revealing governance tensions within AI-powered communication infrastructures. Operational impact is expressed primarily through procedural mechanisms, while limited emphasis is placed on explicit strategic interventions. Socio-political dimensions emerge through repeated references to people, policy and power, highlighting the institutional embeddedness of AI systems. Sentiment analysis indicates mainly neutral evaluations with cautious optimism and critical concern. A quantitative survey of 93 university students from China, Pakistan, and Spain reveals a stronger relationship between algorithmic awareness, perceived bias, narrative manipulation, and democratic resilience in China and Pakistan than in Spain. Overall, the findings demonstrate that AI-mediated systems function not only as technological tools but also as powerful mechanisms of narrative construction, governance, and social regulation, with implications for ethical oversight and democratic resilience.
Artificial Intelligence (AI), Algorithmic Mediation, Hybrid Warfare, Digital Governance; Mixed-Methods; Cross-National Survey
Shang Yan Chen1 Luke Dinh2, 1 St. Margaret’s Episcopal School, 31641 La Novia Ave, San Juan, Capistrano, CA 92675 Moscow, ID USA, 2California State Polytechnic University, Pomona, CA 91768
A high percentage of Americans, over two-thirds of the national population of teens and young adults, feel that they lack a sense of security in themselves. Notably, such individuals often lack daily and consistent practice in confronting complex yet common life dilemmas. Super Youth is the app that precisely targets and reveals each individual’s personality and thinking style when presented with a challenging social scenario that they will encounter occasionally. Flutter, the ChatGPT API, and Firebase Auth all help to generate social scenarios for the user [1][2].By adjusting the app layout to fit within the screen, it can display information concisely on a single screen. A survey was presented to a group of individuals to test how the app helped them improve their social skills. I noticed that most users requested that the app cover more social skills, as they did not feel that it targeted their specific weaknesses.
Social Skills Training, Personality Analysis, AI-Based Education, Mobile Application
Guowei Xiao 1, Ping Wang21 Faculty of Automation Dept, Guangdong University of Technology, Guangzhou, 510006,China 2 Jinqi,Micro Electronics,China
This paper proposes the multirate Convolutional Neural Networks (mCNN) algorithms for an efficient implementation of the 2-Dimensional (2-D) CNN circuits implementation. During the rapid growth in computation power, Deep Learning (DL) using CNN has widened the areas of the Artificial Intelligent (AI) applications. For the layers of the convolution with pooling operation in CNN the work (Franca et al., 1985) has initially applied the multirate algorithms[1-3] to the traditional (non-multirate) convolutional kernel operation of using polyphase architectures resulting in the more efficient implementation of the multirate filtering. In this work we extend it into 2-D CNN by using time-varying coefficient to achieve an efficient implementation with reduced memory(i.e. the line-buffer) size by M-fold(the pooling factor) and the MACs at 1/M of clock running rate. A design example of the first stage of CNN system will be provided. Its results are verified with the Matlab CNN-based digit recognition tool.
CNN, ML, DL, AI, IC, Multirate, 2-D, Signal Processing,DSP, AISC, Filter
Max Liu 1, Garret Washburn21 William P Clements High School, 4200 Elkins Rd, Sugar Land, TX 77479 2 California Baptist University, 8432 Magnolia Ave, Riverside, CA 92504
Teen drivers face disproportionately high crash rates, often due to inexperience and inconsistent attention to basic traffic rules. Safe Student Driving addresses this problem with a multimodal coaching system deployed on both a Raspberry Pi device and a Flutter-based mobile app [1]. The system uses three YOLO-based computer-vision models to detect traffic lights, light-bulb colors, and road signs, an OCR module to read speed-limit values, and an audio model plus IMU data to infer whether turn signals are used during turns [2]. An analysis layer smooths detections over time and triggers prioritized voice prompts through text-to-speech or pre-recorded audio. Key challenges included achieving sufficient model accuracy in varied lighting, running inference fast enough on limited hardware, and designing prompts that inform without distracting the driver [3]. Experiments on sign detection and turn-signal recognition highlight strengths and failure modes, guiding future improvements. Overall, the project demonstrates a practical, low-cost way to help novice drivers build safer habits in real traffic.
Computer vision, Multimodal sensor fusion, Driver behavior monitoring
Ruilai Yang1, Jonathan Sahagun21 Tesoro High School, 1 Tesoro Creek Rd Las Flores, CA 92688 2 California State University, Los Angeles, 5151 State University Dr, Los Angeles, CA 90032
Chess-playing robots represent an ideal testbed for integrating computer vision, robotic manipulation, and artificial intelligence into cohesive autonomous systems [1]. This project addresses the challenge of creating an affordable, accessible chess-playing robot arm capable of competing against human opponents. The proposed solution integrates four core technologies: servo motor control via the Adafruit ServoKit library for precise arm manipulation, inverse kinematics using TinyIK for position-to-angle calculations, OpenCV-based computer vision for chessboard detection and move recognition, and the Stockfish chess engine for AI-powered move computation. Key challenges included achieving reliable board detection under varying lighting conditions, computing accurate joint angles for a 4-DOF arm, and synchronizing physical movements with game state [2]. Experimental evaluation demonstrates 94% board detection accuracy and sub-centimeter positioning precision. The system successfully enables human-robot chess gameplay with configurable difficulty levels, contributing to accessible robotics education and human-robot interaction research.
Robotics, Computer Vision, Human–Robot Interaction, Artificial Intelligence
Noriyuki Suzuki , Nayuta Spiral Works (Independent Research), Japan
Modern cloud and IoT systems increasingly exhibit unstable behaviors caused by non- synchronous data flows, latency fluctuations, and noise amplification across distributed components. Existing stabilization techniques mainly rely on modifying internal system logic or applying domain-specific heuristics, making them difficult to generalize across heterogeneous platforms. This paper introduces an External Emergence-Stabilization Layer (EESL)—a model-agnostic, black-box layer designed to regulate emergent behaviors in distributed systems without requiring internal modifications. EESL operates by monitoring high-level system dynamics and guiding them toward phase-aligned, low-divergence trajectories. We show that many forms of instability in cloud, IoT, and AI systems share a common structure expressed as divergence in behavioral phase, and that stabilizing this phase difference leads to consistent system output. We evaluate EESL under a range of destabilization scenarios, including non-synchronous IoT sensing, latency-induced cloud drift, and perturbed generative behaviors. Across all settings, EESL reduces divergence, shortens recovery time, and improves operational consistency without domain-specific tuning. The results suggest that emergence-level stabilization is a viable and generalizable engineering approach for future distributed and intelligent systems. .
David Kuhlen[0000−0001−8338−7527]1, Technische Hochschule Lubeck – University of Applied Sciences, Monkhofer Weg 239, 23562 Lubeck, Germany
Well-crafted requirement specifications contribute to enabling software developers to implement technical solutions quickly and with high quality. The aim is to contribute to the implementation of good requirements engineering. As the quality of requirements is particularly dependent on compliance with requirement quality criteria, the importance of these criteria is analyzed in relation to predictive success. The most important quality criteria identified were correctness, unambiguity, completeness, and comprehensibility. Moreover, it is possible to improve the quality of requirement specifications in specific quality dimensions through targeted revision. In the course of the RCP/AE experiment, a significant improvement was observed in consistency, correctness, and completeness. Ambiguity was also significantly reduced. Only the improvement in testability was not statistically significant.
Requirements Engineering, Requirement Quality Criteria
George Lu1 1Austin Amakye Ansah2, 1Crean Lutheran High School, 12500 Sand Canyon Ave, Irvine, CA 92618, 1California State Polytechnic University, Pomona, CA 91768
leverages a three-tier architecture, utilizing Flutter for the cross-platform client applications and Google Firebase for backend services, including data persistence, real-time messaging, and authentication [7]. Desktop clients employ platform-specific utilities (WMI on Windows, ps on Unix-based systems) for low-latency process detection, enabling the monitoring of specific applications. Parental control is facilitated through a mobile interface that issues asynchronous commands via Cloud Firestore, allowing for remote process termination [8].Performance benchmarks validate the system efficiency, with the optimized process detection method (tasklist) demonstrating an average latency of 187 ms, and local command execution achieving a mean response time of 5.90 ms. These results confirm that KeepTab provides a responsive, reliable, and scalable framework for modern screen time management.
Screen time, Parental controls, Flutter, Firebase
Rick Galbo, Loyola University Chicago, USA
This paper introduces the Growth-Oriented Neutral Correlation (GONC) Token protocol, a novel construct designed to address the systemic fragility inherent in extant algorithmic stablecoins and the lack of truly uncorrelated assets within decentralized finance (DeFi). Traditional algorithmic stablecoins, such as the mechanism underlying TerraUSD (UST), have demonstrated extreme vulnerability to systemic market shocks, leading to rapid, irreversible failures often termed "death spirals". The GONC architecture mitigates this instability through a three-pronged approach: the Core Stability Mechanism (CSM), the Neutral Correlation Maintenance Model (NCMC), and the Growth-Oriented Reserve Strategy (GORS). The NCMC represents the primary innovation, utilizing a rigorous econometric framework, specifically Dynamic Conditional Correlation (DCC-GARCH), to actively manage and minimize the conditional correlation between the internal reserve assets and the broad cryptocurrency market benchmark. The GORS component integrates yield-bearing, tokenized Real-World Assets (RWAs), providing a stable, exogenous source of capital growth that strengthens the collateral base, moving away from reliance solely on highly volatile crypto-native seigniorage assets. Simulations suggest that by treating correlation as an active risk parameter rather than a static market assumption, GONC significantly enhances resistance to systemic market contagion.
Algorithmic Stablecoin, Dynamic Conditional Correlation (DCC-GARCH), Real-World Assets (RWA), Portfolio Hedging, Systemic Risk, Decentralized Finance (DeFi), Transaction Cost Analysis (TCA)