Contribution: The purpose of this study is: 1) to identify students' early misconceptions in problem understanding and the use of variables/sequencing in a first-year computer programming course using Python; 2) to quantify how early misconceptions evolve over time, and 3) to characterize how misconceptions in the use of variables/sequencing affect learning more advanced computer programming concepts. Background: It has not been clearly established how early misconceptions evolve over time and how they affect learning later in the semester. Research questions: 1) Which are the most common misconceptions in the use of variables/sequencing in a first-year Python course? 2) Do those early misconceptions persist late in the semester? 3) Does the persistence of misconceptions in variable/sequencing hinder progress in learning more advanced computer programming concepts? Methodology: We assessed student performance (n=116) in multiple choice and short answer questions that involved reading and/or writing small flowcharts or snippets of Python code. Findings: Problem understanding remains a challenge for ~45% of students until the end of the semester. Misconceptions in variable assignment/sequencing persist late in the semester for ~20% of students hindering their progress in learning more advanced computer programming concepts.

This study explores innovative strategies to enhance profitability and resilience in Real Estate Investment Trust (REIT) portfolios, addressing limitations in traditional models like ARIMA and GARCH amidst dynamic market conditions. We propose a novel framework integrating Genetic Algorithms (GA) with Machine Learning (ML) techniques-Multiple Linear Regression (MLR), Neural Networks (NN), and Long Short-Term Memory (LSTM) models-leveraging 23 technical and fundamental features. Key contributions include a GA-driven alpha search outperforming market benchmarks, robust ML predictions, and a practical trade execution logic. Tested on a 20year dataset of 150 REITs, our approach surpasses conventional methods, offering a significant advancement in quantitative finance and inviting further exploration.

This paper explores the integration of Artificial Intelligence Generated Content (AIGC) with human-machine intelligence (HMI) to enhance the functionality of Intelligent Transportation Systems (ITS). Adaptive decision-making mechanisms are crucial as transportation networks become increasingly complex, generating vast real-time data from vehicles, infrastructure, and users. AIGC plays a transformative role in optimizing traffic flow through dynamic routing and real-time traffic management. At the same time, human intelligence ensures that these systems remain responsive to ever-changing real-world conditions. In terms of safety, AIGC is employed to simulate complex driving scenarios for autonomous vehicle training and detect traffic anomalies, with human oversight providing contextual decisionmaking in unexpected or ambiguous situations. For sustainability, AIGC develops data-driven strategies to reduce emissions and energy consumption, while human expertise ensures alignment with ethical and environmental goals. This synergy between AIGC and human intelligence is vital for refining real-time decision-making, ensuring both accuracy and adaptability across diverse ITS scenarios. This paper offers a comprehensive literature review on core and supporting AIGC technologies and their applications in key ITS domains. Case studies and initiatives from industry leaders demonstrate practical implementations of AIGC-driven HMI collaboration in ITS. We also address challenges such as compatibility with legacy systems, data privacy, model bias, and scalability. The paper concludes by outlining future research directions, emphasizing the need for scalable, interpretable, and ethically compliant AIGC models. By prioritizing humanmachine intelligence, this work advances the safe, efficient, and sustainable deployment of AIGC-driven HMI solutions in modern transportation networks.

We propose low-cost intelligent reflecting surfaces (IRSs) inkjet-printed on paper substrates to shape electromagnetic (EM) wave-fronts and control reflected wave directions in the millimeter-wave (mmWave) and sub-Terahertz (sub-THz) bands. Unlike conventional IRSs with circuit elements, an FPGA, and a DC supply, our IRSs use inexpensive paper substrates and conductive nanoparticle silver ink, achieving a phase gradient over a 2π period. Simulations and measurements show that our IRSs steer incoming mmWave and sub-THz signals to anomalous reflection angles. For a 0° incidence angle (θi), our 28-GHz-band IRS reflects an incident wave to a 30° anomalous reflection angle, approximately 12 dB higher than a copper plate. Another 100-GHz prototype shows an 8 dB higher anomalous reflection than a copper plate. IRS beamforming efficiency is related to printing resolution and substrate thickness, crucial for realistic wireless communication environments. These IRSs can direct communication signals to non-line-of-sight (NLOS) locations, extending coverage for current and next-generation wireless networks. The low-cost, simple fabrication process supports potential mass production, facilitating widespread adoption in future wireless communication systems like the Internet of Things (IoT) and 6G.

The shift to battery electric vehicles (BEVs) is a key step towards reducing greenhouse gas (GHG) emissions and decreasing dependence on fossil fuels, ultimately achieving Sustainability Development Goals and more specifically, a more sustainable form of transportation. However, while BEVs help eliminate tailpipe emissions, understanding and consequently optimizing their energy efficiency is still an ongoing task, as battery life is often affected by various driving conditions, including driving behavior, terrain and sometimes weather. Traditional linear modeling techniques often lack the ability to capture the non-linear and complex relationships found in real-world scenarios. For this reason, this study introduces a more sophisticated statistical framework for energy consumption profiling for BEVs, through an in-depth exploratory analysis on key influencing factors and the implementation of both a linear and a custom generalized additive model (GAM) for data modeling. The experimental results showcase that the integration of smoothing splines instead of linear regression modeling further refines the proposed model by offering improved accuracy of the order of 20%.

Modeling and simulation is a well-proven approach for conducting what-if analyses of complex scenarios. However, the current societal and technical challenges require increasingly complex models and larger simulation experiments, which calls for more intelligent approaches to simulation and modeling in all aspects of simulation studies. To this end, we believe simulation and modeling can benefit from the recent advancements in machine learning and artificial intelligence (AI) as well as the emerging powerful and pervasive hardware and computing paradigms and systems. In this article, we examine the existing AI techniques and emerging hardware platforms in the context of the modeling and simulation life cycle, broadly comprising the stages of model creation, calibration, and experimentation. We identify key challenges on the path to deeper integration between AI and simulation techniques and outline research directions towards the vision of a higher degree of automation in simulation-supported scientific discovery.

In today's increasingly complex and volatile stock markets, leveraging advanced machine learning and quantitative techniques is becoming indispensable for enhancing trading strategies and optimizing returns. This study introduces a sophisticated Multi-modal framework that combines Deep Reinforcement Learning (DRL) with Algorithmic Trading Signals and Price Forecasts to improve risk-adjusted returns in equity trading. Utilizing the Proximal Policy Optimization (PPO) algorithm within a custom trading environment built on the FinRL library, our approach integrates advanced algorithmic signals-such as moving average crossovers and oscillator divergence-and incorporates enriched price forecasts from Long Short-Term Memory (LSTM) networks. The proposed framework was rigorously evaluated using a diverse set of 29 out of 30 constituent stocks within the Dow Jones Industrial Average (DJI). The empirical results highlight the effectiveness of the Multi-modal DRL approach, demonstrating significant outperformance over traditional benchmarks, with an annualized return of 16.24%, an annualized standard deviation of 17.49%, a Sharpe Ratio of 0.86, and a Sortino Ratio of 1.27. These findings underscore the potential of Multi-modal DRL frameworks to offer consistent, robust performance and contribute to advancing trading strategies in dynamic market environments.

Moment frames are a widely used seismic force-resisting system that relies on the bending deformation of beams to dissipate seismic energy. Although effective, these frames can exhibit vulnerabilities during severe seismic events. To mitigate this, structural fuses, strategically placed in beam-to-column connections, can protect other structural elements from damage. This research explores a novel approach to fuse design, combining the concept of link fuse with weakening techniques and seismic enhancement methods. The goal is to maximize energy dissipation and damage localization within the fuse, minimizing impact on the overall structural system. For this purpose, a reference model with a reduced flange section along with five reduced-web section methods was investigated. The methods are flexural butterfly (with and without sharp corners), shear butterfly, circular, elliptical, and slit-reduced sections. The base model demonstrated the highest energy dissipation capacity. The flexural butterfly fuses, with and without sharp corners, exhibited superior ductility and maximum displacement capacities, respectively. The elliptical fuse proved to be the most resilient, showing minimal degradation and ranking third overall. The slit fuse effectively localized damage, reducing stress on other components.

The increased sophistication and size of current data sets have made conventional analytics methods ineffective and called for effective strategies to work through large data sets. Hybrid deep learning architectures are examined in this paper as a revolutionary approach to Big Data analysis. Due to incorporating features obtained from various deep learning frameworks involving Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and unique Transformer-based architectures, these structures expound improved performances over several analytical tasks. The authors assess the performance of several hybrid models compared to conventional approaches and single deep learning (DL) methods using suitable parameters like accuracy, time of processing, and scalability. The trends are also done in a year-by-year evolution to show the development of technology, whereas comparative bar graphs are used to show the development of capabilities. Outcomes demonstrate that hybrid architectures are superior to customary methods while having superior scalability and functionality across additive datasets. The contribution of this paper is that it provides a critical analysis of the use of hybrid architectures and the implications of their current deployment and evolution for the establishment of the next generation of analytical systems.

In Autonomous Vehicle (AV) technology, the robustness of Traffic Sign Classification (TSC) systems is crucial for ensuring safe navigation. Currently, TSC systems lack dedicated adversarial recovery methods, making them susceptible to adversarial attacks. This study embarks on an innovative path by adapting six established adversarial recovery methods from general image classification (IC) to TSC in AVs. Prompted by the limited availability of TSC-specific adversarial recovery solutions, our research undertakes a comparative analysis of these six methods to evaluate their applicability and effectiveness in the TSC domain. Each method is meticulously adapted and integrated, considering the unique challenges and requirements of TSC in AV environments. Our findings reveal insights into their performance, with the Purifying Variational Autoencoder (PuVAE) method outperforming others, achieving recovery rates of 96.15%, 79.24%, and 60.18% for Chinese, Belgium, and Germany traffic sign datasets, respectively. Additionally, it demonstrated Structural Similarity Index Measure (SSIM) values of 0.58, 0.53, and 0.66, along with recovery times of 0.0009 seconds, 0.0008 seconds, and 0.0008 seconds for the respective datasets. By highlighting effective recovery methods and addressing the unique challenges of adversarial attacks in TSC, this study contributes to improving the safety and resilience of autonomous driving systems. This research not only fills a significant gap in the safety mechanisms of AVs but also paves the way for future exploration and development of more robust and secure TSC systems, capable of effectively countering a wide range of adversarial attacks.

We describe a method for the creation of a compact model for local layout effects (LLE) using pixelated images of the layers of physical layouts as input features. We incorporate these together with electrical measurements of devices on an integrated circuit as input to a neural network. The neural network combines a variational autoencoder (VAE) with a regression network and is used to identify important features of the layout. Then, using a parallel feature neural network to model the effects of those features, we can determine the functional dependence of those effects. Finally, we incorporate those functions into our existing compact model. Because the VAE takes the full design as input, it represents every possible relevant feature. The model visually reveals which features are important, and these visual features can be converted to numerical features. Existing compact models rely on engineers to determine a priori which features are important, but important dimensions on novel devices can easily be missed. We also show that the constraints of the latent space of the VAE allow us to determine during inference whether a particular device geometry is within the domain of the training data. In addition, we describe a process for training the VAE that improves its ability to interpolate more accurately.

Generative Artificial Intelligence (GenAI) has the potential to transform higher education by generating human-like content. The advancement in GenAI has revolutionised several aspects of education, especially subject and assessment design. In this era, it is crucial to design assessments that challenge students and cannot be solved using GenAI tools. This makes it necessary to update the educational content with rapidly evolving technology. The assessment plays a significant role in ensuring the students' learning, as it encourages students to engage actively, leading to the achievement of learning outcomes. The paper intends to determine how effectively GenAI can design a subject, including lectures, labs and assessments, using prompts and custom-based training. This paper aims to elucidate the direction to educators so they can leverage GenAI to create subject content. Additionally, we provided our experiential learning for educators to develop content, highlighting the importance of prompts and fine-tuning to ensure output quality. It has also been observed that expert evaluation is essential for assessing the quality of GenAI-generated materials throughout the content generation process.

Once we learn a reliable gait, we no longer have to consciously control individual muscles in our bodies to walk. Instead, we mainly make decisions on where we want to end up, at what pace and through which path. This high-level locomotion intent is then decoded through our neuromuscular system to generate walking that fits our environment and goals. Estimating the high-level intent may provide the necessary input to wearable robotic devices to adapt to their user's needs. This study investigated methods to estimate a representation of highlevel locomotion intent, the horizontal walking path. We collected fullbody motion capture and bipolar surface electromyography data from 6 subjects during non-steady-state gait. We trained temporal convolutional networks to predict the walking path directly or parametrically with a critically damped trajectory model, using a mixture of muscle and body posture signals. We achieved a mean reconstruction accuracy for a 1-second walking path corresponding to r 2 = 0.89 using a multimodal model. We simultaneously provided estimates for current and desired walking velocities as constrained by the walking path model. Our approach could provide user interfacing in an interpretable and subject-independent format for wearable robotic devices. Moreover, this high-level intent representation is flexible and able to be synthesized in virtual environments, where it can serve as a surrogate for biosignals of simulated intent-driven robotics.

The unprecedented COVID-19 pandemic exposed critical weaknesses in global health management, particularly in resource allocation and demand forecasting. This work introduces a transformative approach to enhance pandemic preparedness through real-time social media analysis. We present CoViNAR, a novel dataset of 14,000 annotated tweets categorized as "Need" or "Availability" of resources (NAR) during COVID-19. The proposed method combines BERTopic for context-aware filtering with advanced machine-learning techniques. The NAR classification model achieved an accuracy of 96.42%, and an F1-score of 96.43% using SVM with DistilBERT embeddings. We demonstrated the success of implementing the NAR classification model by doing a temporal analysis of tweets from the US, UK, and India for the duration of November 2019 to March 2023. The strong correlation between NAR tweet counts and COVID-19 case surges indicates the effectiveness of the proposed method, offering health authorities a powerful, proactive tool for resource management, that can positively transform crisis response strategies. The presented model successfully addresses a crucial gap in the design of early warning systems.

This paper develops a hybrid quantum channel model tailored for finite-regime quantum key distribution (QKD) in free-space satellite quantum communication. The channel model addresses the fading characteristics of satellite-to-ground links while incorporating the dynamics of satellite orbits. By conducting a comprehensive security analysis under finite-size effects and collective attacks, the research bridges the theoretical and practical aspects of satellite-based QKD. Critical parameters, such as reconciliation efficiency, transmission coefficient, detection efficiency, receiver telescope aperture, beam waist, and effective communication time, are systematically investigated to optimize the secret key rate. The study reveals key trade-offs between the influence of satellite altitude on communication windows, noise levels, and synchronization, which significantly impact quantum channel efficiency. The model improves the accuracy of free-space quantum channel evaluations by presenting a hybrid noise framework that combines quantum Poissonian noise with classical additive white Gaussian noise (AWGN). Our numerical results demonstrate the feasibility and robustness of finite-regime QKD for uplink and downlink configurations in various atmospheric conditions, highlighting the superiority of satellite-based quantum communication over conventional networks.

The integration of some functions involving a 2D Gaussian probability integral is reduced to a single integral, thereby greatly facilitating its numerical evaluation. This is surprising because it is well known that a 2D Gaussian probability integral can only be computed using numerical quadrature. Consequently, one might expect that integrating such a function would necessitate double integrals. A numerical example demonstrates the clear advantages of our method. Applications include RF (radio frequency) design engineering coverage probability calculations.

This paper explores the integration of uncertainty frameworks such as Fuzzy, Neutrosophic, and Plithogenic sets into Large Language Models (LLMs) and Natural Language Processing (NLP). We propose novel models, including Large Uncertain Language Models and Natural Uncertain Language Processing, to enhance linguistic representations and processing capabilities. Furthermore, we extend the theoretical foundation of LLMs and NLP by incorporating Hyperstructures and Superhyperstructures, enabling higher-order generalizations and hierarchical modeling. These advancements provide new perspectives for addressing uncertainty and complexity in language understanding and processing. While the paper focuses on theoretical generalizations, practical validation through computational experiments remains an important direction for future work.

Digital Twins (DTs) are dynamic digital representations of physical objects and systems, including their associated processes and environments. Using real-time data analytics, modelling, simulation and 'what-if' scenarios, they can enable valuable understanding and decision-support for managing a system. One potential application of DTs is the management of natural landscapes. However, despite the critical impact of humans on the natural environment, none of the DTs of the natural environment found in the literature include the human systems in that environment. We propose a modular framework for integrating human systems within a DT of the natural environment, to facilitate simulation and modelling of a range of management contexts, objectives and stakeholders across time and space. We then propose four principles as the theoretical basis for modelling the socio-ecological governance systems that underpin DTs of human systems for managing the natural environment. We also provide a use case related to area-wide integrated pest management as an example of the potential application of the proposed framework for collectively managing the natural landscape. The composition and characteristics of this use case as a DT and examples of user engagement at multiple spatial and temporal scales are also described. Finally, we identify some methods such as agent-based modelling, reinforcement learning, and graph neural networks that could be used to incorporate human systems in a DT, along with some examples of their use for similar tasks from the literature. Integrating human systems in DTs of the natural environment will facilitate the development of novel digital decision-support tools that provide stakeholders with different perspectives of the shared natural environment and enable them to understand the impact of various management methods and actions.