Trained human tutors are in huge demand but short supply, making scenario-based training a valuable tool for providing situational experiences to novices. While tutor training through online lessons has shown significant learning gains, human evaluation of real-life performance is time-consuming and costly. Generative AI, particularly large language models (LLMs), offers a promising solution for assessing complex tutor actions, though research is very limited. This present work explores the practical application of generative AI, focusing on prompt engineering techniques to evaluate human tutors in real-life dialogues. We pull a random selection of 50 transcripts consisting of undergraduate remote tutors working with middle-school students on math. We then assess GPT-4's ability to identify and evaluate tutors providing praise to students and reacting to students making math errors. The use of GPT-4 was found to be proficient in identifying these tutoring situations and assessing tutors application of best practices, highlighting its potential for scalable evaluation. This study provides practical demonstrations of prompt engineering, insights into the use of LLMs for assessing the transfer of learning from training to practice, and discusses the ethical and practical considerations when using LLMs for situational judgment.

The increasing number of actions in the real world makes it difficult for traditional deep-learning models to recognize unseen actions. Recently, pretrained contrastive image-based visual-language (I-VL) models have been adapted for efficient â\euroœzero-shotâ\euro? scene understanding, with transformers for temporal modeling. However, the significance of modeling the local spatial context of objects and action environments remains unexplored. In this work, we propose a framework called LoCATe-GAT, comprising a novel Local Context-Aggregating Temporal transformer (LoCATe) and a Graph Attention Network (GAT) that take image and text encodings from a pretrained I-VL model as inputs. Motivated by the observation that object-centric and environmental contexts drive both distinguishability and functional similarity between actions, LoCATe captures multiscale local context using dilated convolutional layers during temporal modeling. Furthermore, the proposed GAT models semantic relationships between classes and achieves a strong synergy with the video embeddings produced by LoCATe. Extensive experiments on two widely-used benchmarks â\euro“ UCF101 and HMDB51 â\euro“ show we achieve state-of-the-art results. Specifically, we obtain absolute gains of 2.8% and 2.3% on these datasets in conventional and 8.6% on UCF101 in generalized zero-shot action recognition settings. Additionally, we gain 18.6% and 5.8% on UCF101 and HMDB51 as per the recent â\euroœTruZeâ\euro? evaluation protocol.Â

The rapid expansion and deployment of AI technologies in various sectors have raised critical concerns about fairness, bias, and ethical implications. Addressing these concerns, the novel integration of structural fairness mechanisms and active learning frameworks into the Mistral Large model presents a significant advancement in creating AI systems that are both highly accurate and equitable. The proposed approach involves enhancing the model architecture with fairness-aware layers and incorporating a modified loss function that penalizes biased predictions, thereby promoting equitable outcomes across diverse demographic groups. Additionally, the implementation of an active learning framework prioritizes the selection of the most informative data points, which not only improves model accuracy but also maintains fairness throughout the learning process. Experimental evaluations demonstrate substantial improvements in performance metrics such as accuracy, precision, recall, and F1-score, alongside significant reductions in bias as measured through disparate impact and equal opportunity difference. The results underscore the feasibility and effectiveness of embedding fairness considerations into the core design and training processes of AI models, paving the way for the development of robust LLMs that adhere to ethical standards and promote social justice. The methodologies and findings provide a comprehensive framework for future research and practical implementations aimed at achieving both excellence and fairness in AI systems.

In response to the growing demand for personalized online shopping experiences, the primary goal of this initiative is to provide users with an engaging platform to customize wrist watches before making purchase decisions. This project introduces an advanced 3D Watch Customization application that seamlessly combines the capabilities of Unity for 3D rendering and for an interactive user interface. Leveraging Unity's powerful 3D rendering, the app allows users to explore an extensive collection of watch models recreated in three-dimensional space. Through an intuitive Unity-based interface, customers can personalize various watch aspects, such as dial design, strap material, and case finish. A standout feature of the app is the incorporation of AR technology, allowing users to visualize their customized watches in real-world contexts. By utilizing device cameras, users can superimpose the virtual watch onto their wrists, offering a realistic preview of how the customized timepiece will appear and fit. This AR functionality seamlessly integrates into the application, enhancing the overall user experience and instilling confidence in the purchase decision.

Grid-tie voltage source converters (VSCs) can operate in three distinct modes: AC-dominant, DC-dominant, and balanced, depending on the placement of the stiff voltage sources. The varying operation modes of the VSCs traditionally demand different synchronization control techniques, posing challenges for the power systems to accommodate and coordinate the VSCs. A promising universal control technique for VSCs is the DC-link voltage synchronization control (DVSC) based on a lead compensator (LC). The LC DVSC stabilizes both the DC and AC voltages of a VSC while achieving synchronization with the AC grid. This results in a dual-port grid-forming (DGFM) characteristic for the VSC. However, there has been very limited study on the stability and synchronization controller design of the VSCs with the LC DVSC operating in various modes. To bridge this gap, the paper presents a quantitative analysis on the stability and steady-state performance of the LC DVSC in all three operation modes of the DGFM VSC. Based on the analysis, the paper provides step-by-step design guidelines for the LC DVSC. Furthermore, the paper uncovers an instability issue related to the LC DVSC when the DGFM VSC operates in the balanced mode. To tackle the instability issue, a virtual resistance control is proposed and integrated with the LC DVSC. Simulation results validate the analysis and demonstrate the effectiveness of the DGFM VSC with LC DVSC designed using the proposed guidelines in all three operation modes. Overall, the paper demonstrates the feasibility of employing the DGFM VSC with the LC DVSC for all three possible operation modes.

This work outlines the procedure to establish a system for assessing materials' complex permittivity and permeability within the 110-170 GHz frequency range. We present the employed methodology and the calibration procedure for the system, incorporating a dual approach using TRL (Thru-Reflect-Line) and GRL (Gated-Reflect-Line) methods. Following that, a smoothing technique is used to enhance the accuracy of the results. Tests were conducted on a HIPS sample to validate the system's performance, demonstrating the results' reliability across the entire measured bandwidth.

This survey paper provides an extensive review of the application of Large Language Models (LLMs) in network operations and management (NO&M). It outlines the transformation in NO&M driven by LLMs, emphasizing their potential to address challenges such as network design, automation, optimization, and security. The paper explores how LLMs can enhance traditional methods by automating complex tasks, improving network agility, and providing scalable and adaptive solutions to emerging network demands. Key contributions include a detailed discussion on LLM-enabled techniques that address issues like real-time adaptability, scalability, security, and intent-based management. The survey categorizes existing research into several key areas and identifies current limitations, such as integration with legacy systems, explainability, data privacy, and scalability. Moreover, it highlights future research directions, including the need for scalable architectures, energy-efficient models, enhanced security protocols, and ethical considerations in deploying AI-driven solutions. The findings are expected to drive innovation and provide valuable insights for researchers and practitioners aiming to leverage LLMs for advanced networking tasks.

An Artificial General Intelligence (AGI) system must intelligently deal with the semantics of the situations it encounters. The flexibility required for intelligent responses is at odds with the fixed parameter spaces of statistical methods or the limitations of the rule-based syntax of the computing paradigms on which systems can now be built. Overcoming these limitations requires philosophical insights from the work of Kant, Wittgenstein, Heidegger, and Gadamer, covering ontology, epistemology, hermeneutics, axiology, and teleology. Specifically, additional teleological computations (to determine the purpose served, independently from the computations responsible for responding appropriately in the world), are necessary to overcome these shortcomings.These insights suggest using a new abstraction to satisfy the necessity for two computational processes to execute in parallel on their respective, related but different representations. The processes must consist of algorithms specifiable a priori, to apply in all possible situations. In this paper, we address some specific difficulties with concepts and show how these difficulties may be overcome using the new abstraction of higher-level perceptions proposed in the Subjective Emergence of Objective Minds (SEOM) model developed earlier.In this paper, we will focus on the representations and teleological computations necessary to give purpose to the actions of autonomous intelligent agents.

Advances in X-ray computed tomography (CT) has enabled soft-tissue X-ray imaging with improved contrast and resolution. As a result, non-destructive X-ray 3D virtual histology is bridging the gap between radiology and histology. Formalin-fixed paraffin embedded (FFPE) blocks are the most abundant archival tissues, and the backbone of histology. Laboratory-based 3D X-ray virtual histology of FFPE blocks is a promising complementary technique to conventional 2D histology. However, there is no generic optimization and consensus on the optimal settings, and associated artifacts with FFPE blocks. Therefore, a comprehensive study on the technical aspects of X-ray 3D virtual histology of FFPE tissue blocks is needed to extract the wealth of information available from them. In this study, laboratory X-ray micro-CT is first optimized with respect to FFPE block positioning, X-ray imaging conditions and artifacts. Then propagation-based imaging (PBI) is discussed and Paganin phase-retrieval as well as Bronnikov-aided correction algorithms are demonstrated for the FFPE blocks. Afterwards, the utility of PBI and its validity domain with considerations of imaging system parameters and FFPE blocks dimensions is investigated based on transport of intensity equation. Then, high-resolution imaging of FFPE blocks based on local CT, and laminography is discussed. In the end, future perspectives are given by considering state-of-the-art micro-CT scanners using liquid-metal-jet sources, large-area detectors, and photon counting detectors. This article provides a starting point for anyone stepping into the world of X-ray 3D virtual histology on FFPE blocks, but also serves as a useful overview for those who might be more experienced in the field.

This letter provides a simulation study on calibration of a 3D LiDAR with respect to a camera, with point-topoint and point-on-plane correspondences for calibration constraints, by extending our previously proposed 2D LiDAR calibration methods. The calibration performance is examined in terms of vertical scan view and resolution, especially in narrow/sparse scan space, in the presence of sensor noise. It contributes to a guidance on a use of various sensors and dataset depending on contexts and applications with different scan view and resolution, and an understanding of the effect of the different constraints on calibration. As a result, the combination of the point and plane constraints yields better performance than either of them.

This study investigates the potential of Virtual Reality (VR) in reducing stress and inducing relaxation, exploring its applicability in trauma and rehabilitation exercises. A calming VR environment featuring nature videos and soothing sounds was developed. ECG signals from 40 participants were recorded under three conditions: resting, artificially induced stress (via mathematical problems, chess games, etc.), and post-VR exposure. HRV metrics were analyzed using machine learning algorithms-SVM, random forest, logistic regression, gradient boosting, and multi-layer perceptron (MLP), with Random Forest achieving the highest accuracy at 58.3 percent. Paired t-tests revealed significant improvements in HRV metrics from the stress phase to the VR phase, indicating VR's effectiveness in stress reduction. Cohen's d effect sizes demonstrated a medium impact of stress on HRV metrics and small to medium improvements post-VR. These findings highlight VR's capability to reduce stress and induce relaxation, supporting its use in trauma therapy and rehabilitation exercises, although the moderate impact suggests the potential need for combined interventions for substantial stress relief.

Exoskeletons aim to enhance human performance and reduce physical fatigue. However, one major challenge for active exoskeletons is the need for a power source. This demand is typically met with batteries, which limit the operational time of the device. A novel solution to this challenge is a design that enables the generation of electricity during motions where the muscles work as brakes, with the energy stored and subsequently returned to assist when the muscles act as motors. This could lead to a fully autonomous exoskeleton. To achieve this goal, a knee exoskeleton design with a direct drive and a novel electronic board was designed and manufactured to capture the energy generated by the wearerâ\euro™s movements and convert it into electrical energy. The harvested energy is stored in a power bank, and, later, the motion is used to power the exoskeleton motor. Further, the device has torque control and can change the assistive profile and magnitude as needed for different assistance scenarios. Sit-to-stand (STS) motion was chosen as a test case for the exoskeleton device. It was found that, during rising (from sit to stand), the exoskeleton provided up to 7.6 Nm and harvested 9.4 J. During lowering (from stand to sit, (it provided up to 10 Nm and was able to return 6.8 J of the harvested energy. Therefore, the cycle efficiency of the exoskeleton system (return divided by harvesting) is 72.3%. The results show that this technology has the potential to revolutionize exoskeletons and reduce the need for external energy sources.

Teleoperated robotic systems have become instrumental in advancing remote healthcare services, especially in tasks that require precision and expert oversight. The advent of cutting-edge telecommunication infrastructures, such as 5G, has amplified interest in these systems, although their full potential remains untapped. This study delves into the effectiveness of teleoperated robotic systems for medicine dispensing, comparing the performance of Wi-Fi and 5G networks in a transnational setup between two cities – Prague and Munich. We focus on the robot’s ability to accurately dispense a predefined volume of a syrup-like substance, simulating a delicate healthcare operation, under the guidance of a distant operator. Our research examines the system’s holistic performance in real-world implementation across diverse scenarios, encompassing varying network states and feedback methods. Two primary feedback scenarios are considered: one incorporating real-time video streaming and another offering explicit quantitative data on the dispensed volume. Using a blend of quantitative and qualitative methods, we aim to determine the influence of network type and feedback on task efficacy and user satisfaction. This study provides insights into the potential and hurdles of deploying teleoperated robotic systems in crucial healthcare contexts, guiding future advancements in this domain, especially in scenarios, where precision and dependability are crucial.

The nascent tokenized assets industry needs to develop standards around asset definition schemas that permit each industry or vertical to create profiles based on the schemas that are specific for addressing the needs of that industry. Machine-readable asset schemas and profiles are a core element of a proper tokenized assets management lifecycle. As metadata artifacts supporting tokenization, they must be managed as part of the lifecycle. A global network of decentralized registries is needed to provide persistence and accessibility to the published asset schemas and profiles. The notion of schemas, profiles, and registries aligns with the asset-referenced tokens (ART) in the recent EU MiCA Regulation on markets in crypto-assets.

Unmanned aerial vehicles (UAVs) can enhance wireless access by dynamically positioning themselves closer to users while maintaining a backhaul connection to cellular base stations. In scenarios where users are geographically dispersed, multiple UAVs can be orchestrated to establish multi-hop integrated access and backhaul (IAB) connections. Traditionally, finding optimal UAV placement has required computationally demanding methods, such as combinatorial optimization or reinforcement learning, which are often impractical for real-time applications due to their complexity and training requirements. Additionally, UAV operators may lack the capability to solve complex optimization problems during live operations. This paper presents a novel framework that leverages large language models (LLMs) for optimizing the placement of multiple UAVs through iterative structured prompting. The proposed method achieves nearoptimal solutions in significantly fewer iterations compared to traditional methods, making it suitable for real-time deployment without extensive mathematical modeling. Through simulations, we demonstrate that this approach can efficiently manage UAV placements and ensure robust network connectivity and service quality. The LLM-driven framework simplifies problem-solving for UAV network operators, paving the way for its application in more complex real-world scenarios.

A novel SiC trench MOSFET with a P-type half-wrapped shield (HW-TMOS) is proposed and analyzed by the TCAD simulator. HW-TMOS could better shield the gate oxide from the high electric field in the off-state by extending the P shield below the gate to the side wall of the trench gate. The breakdown voltage of the device can be enhanced without causing significant degradation of the output characteristics. The peak electric field value of the gate oxide decreases by 64.9% compared with the conventional SiC trench MOSFET, in addition, the breakdown voltage increases by 13.7%. Meanwhile, the capacitance of the HW-TMOS has been optimized by converting the gate-drain capacitance of the trench gate side wall into the gate-source capacitance. The Cgd of HW-TMOS is 67pF/cm2 and the high-frequency FOM of the device is 51 mΩ·nC, which is 80.1% and 36.4% better than the traditional structure. The HW-TMOS has great advantages in reducing switching power due to the low gate-drain capacitance. Overall, the performance of the new structure is significantly improved, which can meet the high requirements of modern electronics for power devices.

With the development of novel display technologies such as 3-D graphical applications, virtual and augmented reality, there are new demands for more efficient and natural multi-degree-of-freedom (DOF) interaction devices. However, traditional rigid touchscreens provide only the 2-D position of the touch point as input information, which cannot meet the needs of high DOF interactions, and many of the existing interaction technologies suffer from the problems of low resolution and lack of information. At the same time, non-rigid interaction interfaces provide deformability for input and have been shown to extend the richness of input vocabulary. This article proposes a new non-rigid input device, OneTip, for single-fingertip human-computer interaction with 6-DOF. In terms of design and manufacture, OneTip employs the bio-inspired design of Skin-On interfaces to mimic the sensitivity of human skin and provide 6-DOF interaction capabilities. In terms of sensing, OneTip uses the visuotactile sensing technique based on the marker displacement method to achieve high-resolution and multi-modal measurements. We propose a novel fingertip pose estimation method based on incipient slip detection, a non-learning algorithm that does not require registration and priori information. Experiments show that OneTip had good 6-D pose estimation accuracy, with RMSEs of translation and rotation not exceeding 0.1mm and 2.6°, respectively, within the linear interval. Extensive experiments were also conducted to explore the application of OneTip in typical virtual manipulation tasks and the possibility of combining it with other interaction devices. This work is intended to serve as a reference for other researchers exploring innovative interaction techniques.

Deep learning (DL) has gained prominence as an effective approach for enhancing the efficiency of various applications including smart grids (SG). Although these models excel significantly in the classification tasks of power quality disturbances, their vulnerability to trojan attacks introduces potential complications. In this paper, we introduce two novel algorithms for executing trojan attacks on DL models handling time series data in SG, tailored for both white-box and black-box. For white-box, our algorithm titled 'Sneaky Spectral Strike (S 3)' utilizes the frequency domain and trigger optimization to perform trojan attacks, which demonstrates a remarkable average fooling rate of 99.9% across various DL models. The algorithm also balances the signal-to-noise ratio, trojan model accuracy on clean data, and fooling rate to be highly effective in fooling DL model and imperceptible to human observers in the power control center (PCC). For black-box, we propose a novel algorithm, 'Lite Datanet Sneaky Spectral Strike', that integrates a simple DL model with a small sample dataset to create trojan triggers that are highly effective, stealthy, and transferable to the DL model deployed in PCC. This approach achieves a 99.86% average fooling rate for different advanced DL models, highlighting the effectiveness of resource-efficient strategies in DL-based SG. Both algorithms underscore the potential vulnerabilities in DL models used in SG , and mark a significant advancement in adversarial machine learning.