Large language models are experiencing a significant surge of attention and rapid development. It is happening mainly due to the publication of OpenAI's ChatGPT models: GPT3.5-turbo and GPT-4. This article uses prompt engineering to present an innovative approach to synthetic data generation and knowledge distillation. Specifically, we focus on three methods: basic prompts, composite prompts, and similarity prompts. This research aims to investigate the potential of these techniques to address the problem of unbalanced datasets, a common issue in machine learning applications. Experimental results reveal that none of the prompt-based strategies achieve scores on par with the entire dataset. However, the similarity prompts method shows promising potential, outperforming other approaches. The study suggests a significant opportunity to develop these techniques further to generate more diverse synthetic data. Although the results are preliminary, they open up exciting possibilities for future research in this area, including integrating more advanced versions of Large Language Models and exploring other machine learning domains.

In conventional 3D printing, the layer-by-layer approach leads to mechanical weaknesses, particularly in the vertical tensile strength (Z-axis) and the shear resistance between layers. To address this, our study introduces a novel non-planar toolpath planning framework for 3-axis printers, named WeaveX, inspired by nature. This method involves moving the nozzle in the XY plane while periodically adjusting its height in the Z-axis, enhancing interlayer bonding and shear resistance. We developed two distinct toolpath designs: Scheme 1, which varies layer thickness, and Scheme 2, which maintains a consistent layer thickness. These designs were closely examined to understand their impact on toolpath width and layer thickness, considering various parameters. Both schemes resulted in "dumbbell"-shaped toolpath geometries, a characteristic that can be lessened by reducing print speed. Mechanical tests revealed that objects printed using these schemes significantly outperform traditional planar toolpath methods in terms of mechanical strength, showing improvements of 31.9% and 67.5% in interlayer shear resistance. Notably, these new strategies can be combined with each other or with conventional methods, broadening their potential applications.

This work reports NbN as a schottky gate on AlGaN/GaN high electron mobility transistor (HEMT). We observed a significant reduction of gate leakage current and three order of magnitude improvement in ON/OFF ratio for NbN gated HEMT as compared to that of Ni/Au gated control HEMT sample. We measured an enhancement of transconductance (maximum) in case of NbN HEMT (85 mS/mm) as compared to control sample (78 mS/mm). The all nitride-HEMT also exhibits the better thermal stability at higher temperatures ( 200°C). The subthreshold slope is also seen to decrease from 266 mV/decade for Ni/Au MSHEMT to 72 mV/decade for NbN MSHEMT. Lastly, our study confirms that the use of NbN as a gate could significantly improve the performance of all-nitride AlGaN/GaN HEMT with better threshold voltage stability and reduced hysteresis in transfer characteristics as compared to the conventional metal gated HEMT.     

This letter considers a reconfigurable intelligent surface (RIS)-aided indoor visible light communication system, where a mirror array-based RIS is deployed to assist the communication from a light-emitting diode (LED) to multiple user terminals (UTs). We aim to maximize the sum-rate in an entire serving period by jointly optimizing the orientation of the RIS reflecting unit, the time fraction for the UT, and the transmit power at the LED, subject to the communication and illumination intensity requirements. To solve this high-dimensional non-convex problem, we transform it as a constrained Markov decision process. Then, a soft actor-critic (SAC)-based deep reinforcement learning algorithm is proposed with the goal of maximizing both the average reward and the expected policy entropy. Simulation results prove the effectiveness of the proposed SAC-based joint optimization design in improving the sum-rate and long-term average reward.

Transformer saturation is a common issue in megawatt converter-based grid emulators (GEs) when emulating grid faults. This problem necessitates the use of anti-saturation control (ASC) with GEs. However, conventional ASC methods tend to distort the emulated grid voltage or even interact with the voltage control (VC) of GEs, causing instability of the system. This paper, thus, proposes an adaptive ASC method and superimposes its output command to both modulation and VC references, which not only alleviate transformer saturation with the minimized voltage distortion, but mitigate its adverse interaction with the VC of converter-based GEs. Experimental results confirm the effectiveness of the adaptive ASC.

In recent years, advancements in Artificial Intelligence (AI) have accelerated, edging us closer to achieving Artificial General Intelligence (AGI). However, alongside these developments, there has been a surge in public concern regarding potential unpredictability and adverse effects of AGI. We hypothesize that the laws of physics, particularly those of Nonequilibrium Thermodynamics, may introduce unexplored constraints on AI. These constraints could be constructively leveraged to predict, monitor and limit potential future paths in the development of AGI. The proposed physics-based approach to understanding AGI includes viewing AGI as a part of a bigger system (the real world), emphasizing the significance of computing costs, establishing links to biological phenomena and applying mathematical tools for rigorous analysis. We hope that this framework will promote an innovative direction in research, adding another powerful method to the toolkit for understanding possible trajectories of AI, alleviating public apprehension and shaping the future discourse of AI research.

The modular multilevel converter (MMC) has recently emerged as a promising candidate for medium-voltage megawatt grid emulators (GEs). However, when emulating grid faults, the MMC-based GE may encounter the transient overvoltage and voltage imbalance issues in its submodule (SM) capacitors. Following the mechanism analysis of these issues, this paper proposes an enhanced SM capacitor energy balancing approach. Differing from traditional methods, this approach effectively mitigates these issues even when emulating faults with residual voltages close to zero, ensuring the reliable operation of SM capacitors within 70%~110% of the rated voltage. Lastly, the effectiveness of the enhanced method is confirmed by the controller hardware-in-the-loop test.

Late-life balance disorders remain a severe problem with fatal consequences. Perturbation-based balance training (PBT), a form of rehabilitation that intentionally introduces small, unpredictable disruptions to an individual's gait cycle, can improve balance. The Tethered Pelvic Assist Device (TPAD) is a cable-driven robotic trainer that applies perturbations to the user's pelvis during treadmill walking. Earlier work showcased improved gait stability and the first evidence of increased cognition acutely. The mobile Tethered Pelvic Assist Device (mTPAD), a portable version of the TPAD, applies perturbations to a pelvic belt via a posterior walker during overground gait, as opposed to treadmill walking. Forty healthy older adults were randomly assigned to a control group (CG, n = 20) without mTPAD PBT or an experimental group (EG, n = 20) with mTPAD PBT for a two-day study. Day 1 consisted of baseline anthropometrics, vitals, and functional and cognitive measurements. Day 2 consisted of training with the mTPAD and post-interventional cognitive and functional measurements. Results revealed that the EG significantly outperformed the CG in several cognitive (SDMT-C and TMT-B) and functional (BBS and 4-Stage Balance: one-foot stand) measurements while showcasing increased confidence in mobility based on FES-I. To our knowledge, our study is the first randomized, large group (n = 40) clinical study exploring new mobile perturbation-based robotic gait training technology.

This paper presents an analytical method for determining the electron mobility in N-polar GaN het-erostructures. The wave function, charge density, and Fermi potential are obtained analytically by combining Airy functions in the channel and spacer regions and utilizing first-order perturbation theory. Importantly, this approach does not rely on any external input such as charge density or electric field. The analytically obtained wave function, charge density, and Fermi potential are then utilized to estimate the scattering rates resulting from various processes in N-polar MIS-HEMTs. Matthiessen's rule is applied to combine the individual scattering rates. The obtained results are validated against experimentally measured mobilities for both conventional GaN and composite GaN-InGaN het-erostructures over a range of channel layer thicknesses.

The bit error rate (BER) analysis of non-orthogonal multiple access (NOMA) has been widely considered in the literature with the assumptions of perfect and imperfect successive interference cancellation (SIC). For both cases, exact closed-form formulas were derived under various channel models, number of users, and modulation orders. However, all the analysis reported overlooked the transformations that affect the probability density function (PDF) of additive white Gaussian noise (AWGN) after the SIC process. Therefore, the signal model after the SIC process is generally inaccurate, which makes the analysis just approximations rather than exact because the noise after SIC is not Gaussian anymore. Therefore, this letter derives the exact noise PDF after the SIC process and evaluates its impact on the BER analysis. The analytical results obtained show that the noise PDF after SIC should be modeled as a truncated Gaussian mixture. Moreover, the PDF after successful and unsuccessful SIC should be modeled differently.

Mobile artificial intelligence-generated content (AIGC) is an innovative technology that can support the evolution and updating processes of virtual twins (VTs) in human digital twin (HDT) systems. With a reliable and efficient automatic information generation process, the requirement for a timely physical-to-virtual synchronization in HDT can be satisfied. While such an AIGC-enabled HDT system can facilitate modelling high fidelity VTs, generating rare disease data and providing timely customized services, it may suffer from poor understanding of contexts, lack of creativity, and various security and privacy concerns. In this paper, we propose a new framework, which integrates federated learning (FL) and semantic communication (SemCom) to enhance performance in such a system while improving accuracy and convergence properties. Such an integrated FL-enhanced SemCom (FLeS) solution, however, comes with its own challenges. First, we present a holistic architectural framework for the proposed FLeS paradigm for mobile AIGC-enabled HDT systems and discuss the associated design requirements and challenges. We later present some key technologies necessary to realize such a solution before elaborating on some technical issues to suggest future directions. We believe that this article will open up new research opportunities and motivate new research efforts toward incorporating FLeS techniques for mobile AIGC, especially in emerging mobile services such as HDT.

This paper proposes a new secure and privacy-preserving prediction-enhanced solution for reliable physical-to-virtual communications in human digital twin (HDT) systems. With such a prediction-enhanced connectivity (PeHDT) framework, the evolution of any virtual twin (VT) could be triggered in real-time or in advance using the expected state of its physical counterpart. This ensures the continuous maintenance of a true replica of each physical twin (PT), thus relieving the need for timely PT-VT synchronization while the VT-experienced delay is reduced to zero or close to zero. We adopted a secured federated multi-task learning technique to meet the security and privacy constraints of HDT and employed a single server discrete-time batch-service queue framework when characterizing the batching process to reduce the communication burden. Furthermore, we introduced a prediction verification framework to improve the performance of the proposed PeHDT framework. The resulting problem was formulated as a constrained Markov decision process and was solved by introducing a primary-dual deep deterministic policy gradient (DDPG) algorithm. Through a joint investigation of communication, batching and prediction verification schemes, the simulation results show that the proposed PeHDT framework can greatly reduce both the VT-experienced delay and the PT-VT communication time without compromising the specific requirements of HDT.

This study presents a passive terahertz (THz) optoelectronic beamforming network (OBFN), utilizing optical dispersion components to induce time delays or phase shifts in signals with varying frequency components. Two feasible schemes are compared and analyzed in terms of implementation difficulty. To validate the concept, feasibility experiments are designed and prototype circuits towards photonic integration are proposed. A comparative analysis of THz beamforming schemes was performed, indicating that the proposed OBFN is beneficial for various THz applications with high beam flexibility and energy efficiency.

Until recently, intensity modulations in synthetic aperture radar (SAR) altimetry waveform tails have been considered a nuisance for geophysical-parameter retrieval. These modulations are actually predictable and might be exploited using a spectral analysis of the waveform tails. After [1], a more elaborated analysis is performed to improve the interpretation of these SAR altimeter spectra. A fast numerical model is developed to explain the modulation mechanisms in focused SAR altimetry waveform tails. Using numerical solutions, standard analytical closed-form solutions, are demonstrated to be invalid to retrieve ocean-wave-spectra retrievals from nadir altimeters. Although not valid, a closed-form derivation provides intuitive insights about the information contained in a SAR altimetry cross-spectrum. Under moderate environmental conditions (significant wave heights of ∼2 m), a closed-form solution might still be useful to infer swell-wave spectra from swath-altimetry SAR spectra, like those of the proposed Sentinel-3 Next Generation Topography mission. Comparable to side-looking SAR ocean processing, the cross-spectral analysis for nadir signals reduces noise and might remove the 180-degree ambiguity of the wave direction. Since the synthetic aperture length of nadir altimeters is larger than sidelooking SARs, sublook processing can be performed to compute multiple cross-spectra for the same scene. With a slightly changing observation geometry, resulting cross-spectra reveal slightly different parts of the ocean-wave spectrum. The resulting cross-spectral stack can thus be used to improve the retrieval of ocean-wave parameters. Retrieved ocean-wave parameters shall then enhance the sampling of the global wave field, but also serve to advance more consistent sea-state-bias corrections.

Power system resilience analysis methods aim to assess the ability of a power system to withstand and recover from high-impact, low-probability events, such as natural disasters. These methods often entail performing AC Optimal Power Flows (OPF) using nonlinear optimization algorithms for numerous sequential Monte Carlo (MC) iterations, which require high computational resources. This may restrict the number of feasible simulations, with the risk of reducing the reliability on the analysis outcomes and limiting the scope of the resilience assessment. This paper introduces the use of Artificial Intelligence (AI) to enhance the computational performance of power system resilience analysis. The paper proposes to use artificial neural networks (ANN) as a surrogate model to approximate the OPF solution for different scenarios and contingencies. The case study demonstrates that ANN provide an interesting solution, reducing the computational burden to a fraction of a classical OPF approaches, while keeping remarkable accuracy in the estimation of resiliency KPIs, such as energy not served (ENS) and survivability. A case study of resiliency analysis with ANN application is presented. The analysis is conducted with reXplan, a novel tool for climate power system resilience analysis, and an ANN built on Pytorch framework. SimBench, an open dataset for network analysis, provides the grid data, generation, and demand profiles. A comparison of the KPIs calculated with ANN and simulated results with OPF is provided, showing the effectiveness and efficiency of the proposed approach.

To facilitate intelligent vehicles in making informed decisions and plans, the precise and efficient prediction of vehicle trajectories is imperative. However, a vehicle's future trajectory is not solely determined by its own historical path; it is also influenced by neighboring vehicles (NVs). Hence, understanding the interactions between vehicles is crucial for trajectory prediction. Additionally, the computational challenges posed by long sequence time-series forecasting (LSTF) add complexity to trajectory prediction tasks. This paper introduces a novel network, named Sparse Attention Graph Convolution Network (SAGCN), designed to comprehensively consider the trajectory interaction details of multiple vehicles, optimizing the LSTF for the target vehicle (TV). Specifically, grounded in real-world driving scenarios and vehicle interaction nuances, a multi-vehicle topology graph is formulated to amalgamate the historical trajectories of the TV and the interaction trajectories of NVs. The SAGCN network employs the Graph Convolutional Network (GCN) to assimilate and analyze diverse features within the multi-vehicle topology graph, subsequently computing the future trajectory of the vehicle through a sparse attention mechanism. The proposed method is validated and evaluated using natural datasets. The results demonstrate that, in comparison to state-of-the-art methods, the SAGCN network presented attains exceptional prediction accuracy and satisfactory time efficiency when predicting the trajectories of TV in LSTF.

The rate of car accidents has been increasing in recent years, resulting in losses in human lives, properties and other financial costs. To address this important issue, an embedded machine learning based system is developed. The system is capable of monitoring road conditions, detecting driving patterns, and identifying aggressive driving behaviors. The system is based on neural networks that are trained on a comprehensive dataset of driving events, driving styles, and road conditions. The system can effectively detect potential risks and help mitigate the frequency and impact of accidents. The primary goal is to ensure the safety of both drivers and vehicles. The process of collecting data involved gathering information on three key road events: normal street and normal drive, speed bumps, circular yellow speed bumps, and three aggressive driving actions, sudden starting, sudden stop, and sudden entry. The gathered data is then processed and analyzed using a machine learning system designed for limited power and memory devices. The developed system resulted in a 91.9% accuracy, 93.6% precision and 92% recall.

In today's data-driven world, quick access to scholarly info is vital. However, current academic search engines face challenges such as restricted text-based searching, uncertainties related to researcher names, absence of contact details, and lack of profile summaries. To mitigate these issues, we introduce ScholarFace, an innovative concept that could transform how we search for academic knowledge. It uses face recognition and language generation technology to spot scholars in photos effortlessly, giving us rich profiles and summaries of their work. It also offers an interactive chat element for users to get more insights about a scholar, reducing online search efforts. We take privacy and ethics issues seriously and ensure ScholarFace complies with rules and regulations. With ScholarFace, we hope to create a smarter, effortless, more connected scholarly world.