Data-driven modelling of non-linear systems with unknown dynamics is a challenging topic. An integral part of this task is the collection of a data set that fully captures the system's behavior in its application scope. The state-of-the-art methods are limited to heuristic input signal generators or model-based optimization methods with strong a priori knowledge assumptions. Currently, there is no approach that simultaneously learns a model, while being end-to-end differentiable w.r.t. a model-independent objective function in order to avoid learning and data set biases. Therefore, the differentiable model predictive excitation (DMPE) algorithm is proposed. The open-sourced, JAXbased algorithm starts with a randomly initialized model that is learnt in parallel to the excitation process from the gathered data. The model is used in a model predictive control fashion in order to maximize the quality of the resulting data set and keep the system within its constraints. The quality of the data set is measured using the Jensen-Shannon divergence based on the joint state-action space. In three numerical application experiments, the algorithm's general applicability and its improved performance in comparison to the current state of the art is demonstrated.

The work describes the creation of the original model of the RPD Module, a component of the Remote PHY Node. This model incorporates GCP/R-DEPI/R-UEPI integration, which has been implemented using cutting-edge packet processing technologies for the enhanced data transmission between CCAP Core and the cable modem via DOCSIS 4.0. The model is open-source and, therefore, does not impose limitations on various applications and improvements for hybrid fiber-coaxial networks. The developed model enhances the performance of processing data packets and improves the reliability of next-generation broadband cable telecommunication networks.

Meta-learning, or "learning to learn", enables machines to acquire general priors with minimal supervision and rapidly adapt to new tasks. Unlike traditional AI methods that approach each task from scratch using a fixed learning algorithm, meta-learning refines the learning algorithm itself through experience across various tasks, enhancing transferability and generalization. This is especially valuable when data collection is difficult or costly, allowing for effective learning from task sequences while reducing the dependency on extensive target domain data. Consequently, meta-learning has emerged as a promising field in machine learning. Although existing surveys provide valuable insights into meta-learning, they often present methods and applications in isolation and lack coverage of the latest advancements. Given the rapid growth of the field, a comprehensive survey is both necessary and challenging. Moreover, meta-learning algorithms often remain disconnected, with no unified framework to explain how they facilitate "learning to learn". This survey seeks to bridge that gap by systematizing meta-learning research, offering a thorough overview of strategies to enhance understanding. Additionally, the paper reviews over thirty representative meta-learning methods across models, tasks, and applications, analyzing their characteristics and challenges. To illustrate method performance, we evaluate more than fifteen models on six problems spanning thirteen scenarios, emphasizing the importance of selecting appropriate meta-learning approaches for practical applications.

The deterministic generation of robust soliton comb has significant meaning for the optical frequency combs to be widely used in various applications. As a novel form of microcomb, Soliton crystal holds the advantages of easy generation, high conversion efficiency, and excellent thermal robustness. Here, we report the turnkey deterministic generation of 'Palmlike' soliton crystal with a free-running scheme. The robustness of the turnkey soliton crystal generation is also investigated in multiple aspects, including the success rate, the thermal robustness, and the long-term stability. The experiment results reveal our turnkey soliton crystal can achieve nearly a 100% success rate with a power variation less than ±1.5 dB over one hundred trials of two samples, is insensitive to thermal effects, and is robust to the environment during four-hour laboratory time.

In the rapidly evolving landscape of wireless communication, securing transmitted information remains a critical challenge. This study provides a comprehensive review of secret key generation (SKG) techniques designed to secure physical layer wireless communication channels. We focus on key generation methods that exploit physical layer characteristics such as channel state information, received signal strength, and bit error rate. The computational complexity of these techniques is also examined, emphasizing the trade-offs between security and efficiency for real-world implementations. The review identifies open challenges such as developing low-cost, resource-efficient systems, managing artificial noise, and addressing the impact of mobility and multiuser massive MIMO systems on SKG rates. Our findings aim to guide the development of more efficient, scalable SKG methods for next-generation wireless networks. Additionally, we propose a method for securely generating identical secret keys between nodes using Cuckoo filters and a Map-Reduce model for frame hashing, ensuring long-term security in IoT environments.

Simulating the fine geometry of Metasurfaces (MTS) structures in a conventional way is a difficult task requiring huge computational resources. On the other hand, the metallization can usually be modeled as a (cascaded) impedance sheet(s) with modulation scale of the order of the operating wavelength. Such modeling, while homogenizing the currents and fields on the surface, is accurate in predicting the far-field as well as the near-field down to a fraction of wavelength from the surface. However, analyzing an impedance sheet lying on a grounded slab with the Method of Moments (MoM) often leads to ill-conditioning; it is certainly the case when the range of impedance spans both the capacitive and inductive domains. Such impedance range is in practice required for Reflective Intelligent Surfaces (RIS), and for some MTS antennas. This paper proposes a preconditioner aiming to solve ill-conditioning issues caused by the wide range of the surface impedance. The preconditioner involves a multiplication by the conjugate of the MoM matrix followed by a block diagonal preconditioner. The preconditioning is implemented in an accelerated scheme relying on FFTs. Given the resulting well-conditioned system of equations matrix, the unknown current distribution is efficiently computed iteratively using the Generalized Minimal RESidual (GMRES) method. A fast convergence is observed independently of the impedance range.This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible  

The pursuit for industrial energy efficiency and the mitigation of greenhouse gas emissions motivate researchers to develop or restructure energy materials that can lower the global energy carbon footprint. This study hybridizes three basic Triply periodic minimal surface topologies using into one shape. A novel heat exchanger is proposed by hybridization of 3 basic topologies; Schoen Gyroid, Schwarz Diamond and Fischer Koch S to leverage and combine their individual thermo hydraulic properties. The aluminium alloy used is a thin walled AlSi10Mg. Hybridization was achieved by level set trigonometric functions and modeled by MSLattice. The new shape was simulated as an evaporator using Simulink, MATLAB 2020a running the Tetrafluoroethane (R134a) on the material with a geothermal (sustainable energy) heat source. The highest surface area density is 29.59 l/mm with a wall thickness of 0.12 mm. The best obtained steady state values of specific enthalpy and heat flow rates are 439.4 kJ/kg and 0.34 kW/s respectively. This novelty sheds some light into the 2 phase 3 zone evaporator’s zone length and mass fraction of phase transformation and zonal fluid mass along the volume length, promising better thermo hydraulic results as more Triply periodic minimal surfaces are hybridized.

Large language models (LLMs) have seen rapid advancements in their ability to integrate and generate knowledge from various sources. However, the challenge of efficiently combining information across different modalities-such as text, images, and audio-persists. In this work, we introduce the Iterative Retrieval Augmentation for Multi-Modal Knowledge Integration and Generation (IRAMIG) framework, designed to enhance the integration of multi-modal knowledge. By employing an iterative retrieval mechanism, IRAMIG allows for the dynamic fetching of relevant data informed by context from previous queries. The architecture comprises modality-specific retrieval modules and a fusion component that synthesizes knowledge, ultimately enriching the model's understanding and output generation. Experiments conducted across image captioning and multi-modal question answering tasks reveal significant performance improvements when compared to existing methods, showcasing IRAMIG's potential to refine knowledge integration and enhance the quality of generation in complex multi-modal scenarios.

We propose an artificial intelligence (AI) modeling method that utilizes an arbitrary voxelized mesh of a small PCB via structure, without relying on parametric geometry. Our AI model, based on a multilayer convolutional neural network (CNN), significantly reduces simulation time for small PCB via structures. By employing advanced AI techniques, we enhance the accuracy and quality of predictions while substantially reducing the size of the neural network model compared to our previous work using a fully connected neural network (FCN). The CNN-based model predicts 2-port scattering parameters with a median Huber error of 0.16%, achieving computational speeds over 13,000 times faster than traditional full-wave field solvers.

A radially periodic 2-D leaky-wave antenna (LWA) with high gain at broadside, reduced side lobes, and suppressed open stopband (OSB) is presented in this communication. The antenna is low profile and is defined by a printed, double microstrip bull-eye aperture. To support polarization reconfigurability and application to full-duplex systems, dual-polarization scenarios, etc., the antenna is fed with a compact multi-slot array at the centre of the LWA and this is enabled by substrate integrated waveguide (SIW) technology. When considering differential excitation, the simulations and measurements demonstrate a maximum realized gain of more than 17 dBi at broadside at about 18.4 GHz. Also, thanks to the suppression of the OSB, persistent broadside radiation is supported from 18 GHz to 19 GHz as well as continuous and sustained directive radiation when the main beam scans through broadside, which is in agreement with leaky-wave theory. Applications include satellite connectivity, polarization diversity scenarios, radar and monopulse systems, as well as 5G/6G wireless communications.

Maintaining efficiency, lowering operating expenses, and satisfying growing customer needs all depend on the optimization of logistics within worldwide supply chains. Traditional logistics management systems are proving inadequate as global supply chains get increasingly complicated from events including changing demand, geopolitical disturbances, and legislative differences. With an eye on demand forecasting, route optimization, and inventory management, this work explores how machine learning (ML) technology might be used to maximize logistics processes. A subset of artificial intelligence (AI), machine learning provides sophisticated features for pattern recognition, huge dataset analysis, and real-time decision-making enablement. Logistics can be much improved by means of supervised learning methods for demand forecasting, reinforcement learning for dynamic route optimization, and unsupervised learning for inventory clustering. Moreover, predictive analytics is investigated for its ability to reduce risks by projecting supply chain disturbances depending on outside variables including political events, climate change, and industry trends. The approach used in this study consists in a survey of current literature on machine learning applications in logistics as well as case studies of actual implementations by multinational companies like Amazon and DHL. These case studies show how greatly cost reductions, lower delivery times, and enhanced operational efficiency resulting from machine learning-driven logistics systems have resulted. By constantly changing paths depending on traffic and environmental circumstances, machine learning algorithms used to route optimization have enabled speedier delivery and reduced fuel usage. Machine learning models have exceeded conventional statistical approaches in demand forecasting, therefore enabling improved inventory control and a decrease in stockouts and overstocks. The report also emphasizes the difficulties of application even as the benefits of adding machine learning into logistics are obvious. Widespread adoption is seriously hampered by issues including data protection, the integration of machine learning algorithms with older systems, and the need of specialized knowledge. Still, machine learning appeals as a solution for enhancing worldwide supply chain logistics because of possible cost savings, higher efficiency, and improved customer service. This paper ends with recommendations for next research prospects in the evolution of increasingly sophisticated machine learning techniques fit for the increasing complexity of worldwide supply networks. Furthermore open for investigation are the ethical ramifications of machine learning in logistics, especially with relation to automated decision-making and data protection. Machine learning will become ever more important in driving innovation and optimization in logistics as supply networks change.

This study compares the effectiveness and correctness of grammar error repair with GPT-3.5 to traditional tools. To improve writing quality and communication clarity in a variety of academic and professional contexts, grammar correction is essential. Using assessment measures including accuracy, precision, recall, and F1 score, the study compares the performance of GPT-3.5 with other grammatical correction tools for corrective tasks. In order to evaluate the practical usefulness of the corrections, the data also contains user input and blind testing. The results show that GPT-3.5 outperforms traditional tools in contextual error management and user satisfaction by fixing grammatical errors exceptionally well, even in complicated sentences with contextual nuances. It does, however, have several drawbacks, including as sensitivity to changes in input and sporadic overcorrection. The study recommends enhancing contextual learning and decreasing overcorrection to further enhance algorithms. It also looks at how GPT-3.5 might be used to improve writing assistance and language-learning materials, outlining potential future research areas to maximize grammar correction AI models.

Morphological operators are crucial in image analysis. Their integration into deep learning pipelines could improve performances by extracting or enhancing important image features, either within network architectures or loss functions. However, the difficulties in rendering those operators differentiable hinder their integration. In this paper, we present SoftMorph, a novel framework designed to convert any binary morphological operator defined as a Boolean expression into its differentiable and probabilistic counterpart, compatible with gradient-based optimization. Specifically, we define probabilistic operators as the expectation of the binary operator with respect to the probability of generating each binary configuration. This expectation can be computed trivially from the truth table of the binary morphological filter, as a multi-linear polynomial function. Moreover, we approximate the probabilistic operators with quasi-probabilistic operators directly translated from the Boolean expressions leveraging Fuzzy logic. These quasi-probabilistic operators therefore maintain the computational complexity of the original binary operator. We demonstrate the efficiency and reliability of our method through validation experiments, and evaluate the backpropagation capability of the proposed operators. Finally, we showcase several applications of morphological operators integrated into neural networks for image segmentation tasks.

Transformer-based Large Language Models (LLMs) are progressively entering specialized areas, including the management of optical networks. These models have the ability to automate critical tasks, optimize network performance, and interact with operators, transforming how decisions are made in real-time network environments. This article explores the transformative role of LLMs in optical networks, addressing their application in network design, failure prediction, reconfiguration, and quality of transmission (QoT) estimation. Furthermore, the integration of LLMs in operator-driven decision-making processes is examined, highlighting how these models can provide intelligent recommendations that enhance operational efficiency. Key challenges such as real-time processing, hallucination risks, and domain-specific model adaptation are also discussed, alongside the deployment strategies for LLMs in cloud, edge, and hybrid network architectures. The article concludes by exploring future directions in LLM-assisted optical network management, with insights on human-in-the-loop systems and continuous learning.

This article investigates the impact of ultra-wideband (UWB) electromagnetic interference (EMI) on the performance of a passive frequency doubler. The doubler, fabricated using 0.5 μm GaAs Schottky technology, was characterized under various UWB pulse amplitudes. Experimental measurements of conversion gain and harmonic levels were conducted and verified through simulations. At lower UWB EMI amplitudes, the doubler exhibited reversible performance degradation. However, at higher amplitudes, irreversible damage occurred. The threshold for irreversible damage was experimentally determined. Additionally, an N-norm analysis was performed to assess the cumulative effect of multiple UWB pulse exposures. The feature selective validation method was applied to the second harmonic performance, confirming good agreement between the simulation and measurement results. This research provides valuable insights into the vulnerability of passive frequency multipliers to UWB EMI and contributes to the development of robust protection strategies for critical RF components.

Anthropomorphic phantoms are critical in computed tomography (CT) imaging, simulating human anatomy for image quality and dosimetry evaluation. Traditional phantoms, however, often fail to accurately replicate complex tissue patterns, particularly for advanced technologies like spectral photon-counting CT (SPCCT). This study introduces a framework for creating 3D-printed anthropomorphic phantoms optimized for SPCCT, beginning with a breast phantom. An in-house optimization algorithm was developed to generate an optimal image acquisition protocol, capable of distinguishing between adipose, fibroglandular, skin, and carcinoma tissues. Using the Mars Microlab 5×120 SPCCT system, it was found that a tube voltage of 80 kVp and narrow energy bins 7-27 keV, 27-32 keV, 32-37 keV, 37-42 keV, and 42-80 keV, offered superior results. Various photopolymerizable resins and thermoplastic filaments were tested as tissue-mimicking materials (TMMs), comparing their linear attenuation coefficients with those of the tissues and reference TMMs. The breast phantom was created by segmenting an MRI image from the Duke Breast Cancer MRI dataset into two models: one for skin, fibroglandular, and carcinoma tissues, 3D-printed with e-PLA resin, and another for adipose tissue, replicated using paraffin wax. Multi-energy images of the phantom were then evaluated using both narrow and wide energy bins for validation. While this study focused on a breast phantom, the framework can be applied to create various anthropomorphic phantoms, enhancing SPCCT imaging and potentially improving diagnostic accuracy and patient outcomes.

This review paper explores advanced methods to prompt Large Language Models (LLMs) into generating objectionable or unintended behaviors through adversarial prompt injection attacks. We examine a series of novel projects like HOUYI, Robustly Aligned LLM (RA-LLM), StruQ, and Virtual Prompt Injection, that compel LLMs to produce affirmative responses to harmful queries. Additionally, the paper investigates the robustness of these attacks across different models and prompts. Several new benchmarks, such as PromptBench, AdvBench, AttackEval, INJECAGENT, and RobustnessSuite, have been created to evaluate the overall performance and resilience of LLMs against adversarial prompt injection attacks. Results show significant success rates in misleading models like Vicuna-7B, Llama-2-7B-Chat, GPT-3.5, and GPT-4. The review highlights limitations in existing defense mechanisms and proposes future directions for enhancing LLM alignment and safety mechanisms such as LLM SELF DEFENSE, Unlike previous studies, this paper provides a more comprehensive evaluation by integrating a broader range of attack methods and target models. Our study shows that there is a need for improved robustness in LLMs, which will potentially shape the future of AI-driven applications and security protocols.

Airport operational efficiency is significantly enhanced by the accurate prediction of meteorological conditions. It enables safe and effective air traffic control both en-route and on the ground. Aerodrome systems demand continuous operation, which means that they regularly generate a lot of data thus verifying a forecast's value, consistency and quality is the aim of forecast verification. If a forecast accurately predicts the observed conditions based on objective or subjective criteria, it is considered high quality. A forecast's usefulness in aviation is crucial since it directly affects operational regularity, economy and safety. When the predicted value is within predetermined bounds, the forecast is validated as "correct" or otherwise "incorrect". At present, real time verification is conducted for every component of TAF, TREND forecast, takeoff forecast, area/local forecast and aerodrome warnings and light aircraft warnings. This study assesses the accuracy of the performance-based analysis of the forecasting system of the India Meteorological Department for aviation meteorology as verified with the Meteorological Aerodrome Report (METAR) in conformity with the ICAO's prescribed standards. The evaluation is done over the period from January 2018 to December 2023. The purpose of the analysis is to help duty officers of aerodrome weather forecasts by providing them with insights from previous operations.