The consumption of specialty coffee has increased around the world. Specialty coffee is free from impurities and defects. Specialty coffee is produced in smaller quantities, and its production process is hard and expensive. Traditional coffee beans have defects that affect the flavor of the coffee. It is essential to select the type of coffee to guarantee the best cost and quality. When a human manually classifies the type of coffee, there may be interference due to the human condition. This process has the disadvantage of being subjective. Few studies used Machine Learning methods for predicting specialty coffee classification by analyzing images. This article proposes a framework for classifying specialty coffees by applying Tiny Machine Learning techniques. We developed a model that can help accurately analyze and classify the coffee process and inform the quality of coffee without human interference. We achieved 100% accuracy, and we believe our model can be used effectively in the coffee industry.

This study presents a compact, on-chip Analog Probe Module (APM) to augment the IEEE 1149.4 (or P1687.2) standard, for efficiently probing multiple internal nodes. The complete approach to APM implementation from conceptualization to practical application is discussed in detail. The APM aims to utilize minimum area for a maximum number of probe channels, achieving an optimal size of 4:15. At the transistor level, the design minimizes the impact of glitches in asynchronous operations through a symmetrical layout and a unique arrangement of all probe channels. However, glitches in asynchronous circuits can still exist; hence, a state transition matrix (STM) concept is devised. STMs help visualize hazardous transitions, allowing the identification of a common hazard-free transition sequence suitable for hardware implementation. The verified APM design is integrated with several analog/RF circuits fabricated in a commercially available 0.18µm RF-CMOS process as part of a radar-on-chip system. An important APM application of enabling the prediction of an IC’s corner disposition by measuring DC-node voltages during post-silicon IC testing, is demonstrated for six fabricated ICs.

For homogeneous traffic, where all vehicles are the same type, the traffic state is characterised by speed, flow, density, queue length, etc. In mixed traffic conditions, variations in static and kinematic characteristics among vehicles and the resulting asymmetric interactions that arise, these state variables are inadequate to represent class-wise behaviours. This paper proposes a novel framework for characterising mixed traffic conditions based on vehicle class-wise speeds rather than a single value of the aggregated stream speed. Also, it proposes an area occupancy-based approach to estimate class-wise speeds from class-agnostic disaggregated traveltime data. The empirical validation of the proposed traffic state definitions demonstrates their generalisability. Finally, parametric and non-parametric prediction models are also developed for state and class-wise speed predictions. The empirical results demonstrate that the joint prediction approach (simultaneous prediction of multiple classes using the proposed state definition) is more accurate, computationally effective, and more efficient for practical applications than the marginal predictions using class-wise speed predictions. Moreover, the order of the class-wise speeds is more robustly preserved in the former than in the latter. This research can open doors for a new family of class-wise speed-based traffic management strategies and applications for mixed traffic conditions.

With the rapid development of artificial intelligence technology, the field of reinforcement learning has continuously achieved breakthroughs in both theory and practice. However, traditional reinforcement learning algorithms often entail high energy consumption during interactions with the environment. Spiking Neural Network (SNN), with their low energy consumption characteristics and performance comparable to deep neural networks, have garnered widespread attention. To reduce the energy consumption of practical applications of reinforcement learning, researchers have successively proposed the Pop-SAN and MDC-SAN algorithms. Nonetheless, these algorithms use rectangular functions to approximate the spike network during the training process, resulting in low sensitivity, thus indicating room for improvement in the training effectiveness of SNN. Based on this, we propose a trapezoidal approximation gradient method to replace the spike network, which not only preserves the original stable learning state but also enhances the model's adaptability and response sensitivity under various signal dynamics. Simulation results show that the improved algorithm, using the trapezoidal approximation gradient to replace the spike network, achieves better convergence speed and performance compared to the original algorithm and demonstrates good training stability.

The development of Digital Twins (DTs) represents a transformative advance for simulating and optimizing complex systems in a controlled digital space. Despite their potential, the challenge of constructing DTs that accurately replicate and predict the dynamics of real-world systems remains substantial. This paper introduces an intelligent framework for the construction and evaluation of DTs, specifically designed to enhance the accuracy and utility of DTs in testing algorithmic performance. We propose a novel construction methodology that integrates deep learning-based policy gradient techniques to dynamically tune the DT parameters, ensuring high fidelity in the digital replication of physical systems. Moreover, the Mean STate Error (MSTE) is proposed as a robust metric for evaluating the performance of algorithms within these digital space. The efficacy of our framework is demonstrated through extensive simulations that show our DT not only accurately mirrors the physical reality but also provides a reliable platform for algorithm evaluation. This work lays a foundation for future research into DT technologies, highlighting pathways for both theoretical enhancements and practical implementations in various industries.

This study examines the difficulties with digital accessibility that people who are functionally illiterate or semiliterate in the plumbing and electrical services sector experience, and it suggests a user-centered approach to improve digital inclusion. The goal of the project is to create a personalized web interface that can be used by people with low reading levels to interact with digital devices. The platform attempts to empower service providers and consumers alike by utilizing proactive instruction, graphic iconography, and text written in the local language. High levels of user satisfaction were found during usability testing and feedback gathering, demonstrating the value of customized design techniques in meeting a range of user demands. The study emphasizes the value of user-centric techniques for socioeconomic empowerment, which adds insights to the field of inclusive interface design. The platform is optimized in response to user feedback and its technical implementations. This might have an impact on digital inclusion in a variety of sectors and geographical areas. In the end, the study emphasizes how accessible interfaces that cater to users with different literacy levels may help technology promote inclusion and equality in the world of internet and technology.

In response to escalating energy demand and environmental concerns, renewable energy sources have emerged as crucial alternatives to conventional power generation. This paper proposes an innovative solution to address reliability and performance challenges in renewable hybrid PV and Sustainable Energy Generating Pad (SEGP) energy systems by integrating power electronics devices connected to the AC-DC microgrid system. The study uses ETAP software for simulation to identify Busbar instability and the system's performance. Subsequent adjustments to the power electronics devices result in notable system reliability and performance enhancements of about 98-100%, leading to reduced power losses. These findings underscore the transformative potential of this approach in bolstering societal sustainability by sustaining renewable electricity generation and mitigating environmental impacts. The research significantly contributes to ongoing efforts to bridge the gap between energy supply and demand for stable island mode microgrid operation mechanisms, addressing challenges associated with renewable energy generation with alternative energy source methods.

Forward adaptive transform coding of images requires a codebook of transform matrices from which the best transform can be chosen for each macroblock. Codebook construction is a problem of designing a quantizer for Karhunen-Lóeve transform (KLT) matrices estimated from sample image blocks. We present a novel method for KLT matrix quantization based on a finite-lattice non-causal homogeneous Gauss-Markov random field (GMRF) model with asymmetric Neumann boundary conditions for blocks in natural images. The matrix quantization problem is solved in the GMRF parameter space, simplifying the harder problem of quantizing a large matrix subject to an orthonormality constraint to a low-dimensional vector quantization problem. Typically used GMRF parameter estimation methods such as maximum-likelihood (ML) do not necessarily maximize the coding performance of the resulting transform matrices. To this end we propose a method for GMRF parameter estimation from sample image data, which maximizes the high-rate transform coding gain. We also investigate the application of GMRF-based transforms to variable block-size adaptive transform coding.

This sign language becomes an integral part of communicating with deaf people who are mute. otherwise, it is a long way barrier to interacting with other people who are unfamiliar with this way of communication. So the sign language recognition system is a way to remove this long-way barrier to better and proper communication. As there are quite several sign languages in use across the globe, it is not practically possible for everyone to learn these languages. This paper proposes the technique derived from deep learning to sign language recognition using custom datasets for two different languages. Each dataset has 35 classes with each class having 1200 different images. Hereby, in this work, we further present a process of describing the datasets of ISL and ASL in this respect by image classes followed by labeling the same. The TensorFlow model is trained with these datasets to achieve a system to identify sign language. High accuracy can still be obtained even though the dataset size is small.

Speed regulation of the electro-mechanically actuated systems is common across broad application segments, including industrial and residential sectors. The growth in demand for efficient energy conversion has made permanent magnet synchronous machines (PMSM) a friend, rather than a foe. Vector controlled PMSMs in speed regulation mode being the dominant form of control, accurate speed, and position measurement is crucial. However, a misalignment in the position sensor with respect to the true rotor zero position in speed regulated PMSM drives leads to inefficient energy conversion with the possibility of unstable motor behavior. The impact of such position sensor fault on the overall system performance is presented from an efficiency as well as a control system point of view. The theory presented is further corroborated with simulation and experimental results to practically demonstrate the importance of the position sensor misalignment fault from a speed regulated application point of view. A strategy to quantify the severity of the fault is also proposed and validated with experimental results, supporting fault mitigation activities.

Information Communication and Technology (ICT) accounts for an increasing share of global Green House Gas (GHG) emissions. Wireless circuits and systems are indispensable in across all ICT sectors, from cellular networks through satellite communications, to the Internet of Things (IoT). While Life Cycle Assessment (LCA) is an industry-standard methodology for assessing the environmental impact of systems, there has been no comprehensive LCA study focusing on RF systems. We present the first comparative LCA specific to RF and microwave applications, and a design-for-sustainability guide covering mainstream RF applications. With Integrated Circuits (ICs) having the largest environmental impact, the trends in RFICs and MMICs are summarized and evaluated from an environmental perspective. Moving to mmWave frequencies, beyond 20 GHz, results in a shift to over 50% smaller node size above 28 GHz, but with minimal reduction in the core chip area, which inevitably increases the environmental footprint of 5G/6G mmWave systems. We then review active and passive microwave circuits focusing on phased array elements and filters, both distributed and lumped. A bespoke model for RF PCBs is also presented to model the surface finish and transmission line or antenna area accurately. Our LCA indicates that design choices such as the CMOS process, PCB material and surface finish, can have a large environmental impact at the manufacturing stage. We highlight the importance of low-loss components, comparing microstrip and LC filters, where a higher end-to-end RF system efficiency directly translates to a lower global warming potential (GWP), reducing Scope 2 GHG emissions.

In this paper, we introduce an enhanced textual adversarial attack method, known as Saliency Attention and Semantic Similarity driven adversarial Perturbation (SASSP). The proposed scheme is designed to improve the effectiveness of contextual perturbations by integrating saliency, attention, and semantic similarity. Traditional adversarial attack methods often struggle to maintain semantic consistency and coherence while effectively deceiving target models. Our proposed approach addresses these challenges by incorporating a threepronged strategy for word selection and perturbation. First, we utilize a saliencybased word selection to prioritize words for modification based on their importance to the model's prediction. Second, attention mechanisms are employed to focus perturbations on contextually significant words, enhancing the attack's efficacy. Finally, an advanced semantic similarity-checking method is employed that includes embedding-based similarity and paraphrase detection. By leveraging models like Sentence-BERT for embedding similarity and fine-tuned paraphrase detection models from the Sentence Transformers library, the scheme ensures that the perturbed text remains contextually appropriate and semantically consistent with the original. Empirical evaluations demonstrate that SASSP generates adversarial examples that not only maintain high semantic fidelity but also effectively deceive state-of-the-art natural language processing models. Moreover, in comparison to the original scheme of contextual perturbation CLARE, SASSP has yielded a higher attack success rate and lower word perturbation rate

Accurate estimation of AC losses is crucial to successfully implement high-temperature superconductors (HTS) in the armature of electric machines. In this estimation, a purely sinusoidal current is often assumed. However, the connection to power electronic converters will introduce harmonics in the transport current, generally leading to higher losses. In this paper, four separate distorted current waveforms produced by space vector modulation were implemented to assess how harmonics from power electronics influence the HTS AC losses in a 2.5 MW superconducting aviation motor. By varying the switching frequency between 5 and 40 kHz, the amount of harmonic distortion was varied between the four cases. The ReBCO and copper layers were both modeled using a multilayer H-A-formulated model. While the harmonic distortion was found to have a minor impact on ReBCO layer hysteresis losses, it increased eddy current losses in the copper stabilizer by between 250 % and 471 %, depending on the amount of distortion. Consequently, the total losses increased by between 60 % and 123 %, with a higher copper loss fraction.

In today's scientific landscape, Directed Acyclic Graphs (DAGs) are pivotal for representing task dependencies in data-intensive applications. Traditionally, two dominant bottomup DAG scheduling approaches exist: one overlooks communication contention and the other fails to exploit parallelization for improving latency. This study distinguishes itself by advocating a top-down approach prioritizing latency or cost optimization in multi-tier environments to fulfill QoS and SLA requirements. Our strategy effectively mitigates bandwidth contention and facilitates parallel executions, leading to substantial completion time reductions. Our findings suggest that myopic knowledgebased scheduling, emphasizing latency or cost minimization, can yield benefits comparable to its look-ahead counterparts. Through latency-efficient and cost-efficient topological sorting, our wDAGSplit strategy introduces a two-stage partitioning and scheduling approach. Its simplicity and adaptability extend its usability to DAGs of any scale. Evaluated on over 100,000 real-world DAG applications, wDAGSplit demonstrates latency enhancements of up to 80x compared to Edge-only scenarios, 15x to Near-Edge-only, and 6x to Cloud-only. In terms of cost, our approach achieves enhancements of up to 60x compared to Edgeonly scenarios, 250x to NE-only, and 70x to Cloud-only. Moreover, for DAGs with 50 tasks, we achieve 5x reduced latency compared to previous approaches, along with a complexity reduction of up to 24 times.

The rapid advancement towards 6G networks is set to revolutionize media services, offering unprecedented capabilities such as high-fidelity augmented reality, virtual reality, and ultra-high-definition video streaming. However, these advancements necessitate superior network management techniques to handle increased bandwidth demands and stringent latency requirements effectively. This paper introduces the concept of network digital twins as a pivotal solution in managing and optimizing 6G media services. By creating a virtual replica of the network that can predict and simulate real-world behaviors, digital twins facilitate proactive optimizations and real-time decision-making. We present a comprehensive simulation study demonstrating how network digital twins can significantly enhance bandwidth allocation, reduce latency, and improve overall media service quality in a 6G environment. Our findings suggest that digital twins not only promise to improve operational efficiencies but also enhance the end-user experience in nextgeneration media applications.

Electromagneto-quasistatic (EMQS) field formulations allow to model resistive, capacitive, and inductive field effects while neglecting wave propagation. These field formulations are based on the Darwin-Ampére equation and yield different approximations of the full set of Maxwell's equations depending on the choice of additional equations. Various discrete EMQS formulations are analyzed using the port-Hamiltonian system framework. It is shown that several symmetric EMQS formulations, e.g. combinations of the Darwin-Ampére equation and the Maxwell continuity equation, yield port-Hamiltonian differential-algebraic system of equations (pH-DAE) which implies their numerical stability, energy conservation related to a specific EMQS variant of the Hamiltonian and dissipativity results.

A fundamental radiation limit for quasi-isotropic antennas is rigorously presented utilizing vector spherical modes within the framework of a mode-matching technique. The theoretical derivation is substantiated through diverse numerical analyses demonstrating that quasi-isotropic radiation employing an omnidirectional antenna is realizable with unity gain across nearly all angles and a maximal gain of 1.1929 (0.7660 dBi) at specific finite angles attributable to the Gibbs phenomenon. Practical implementation necessitates the introduction of a linear array comprising traditional element antennas such as Hertzian dipoles, half-wave dipole antennas, and rectangular patch antennas. Our array configuration is validated using commercial software packages, and the quasi-isotropic radiation patterns they generate exhibit favorable agreement. The ratio between the numbers of element antennas and effective Legendre modes governs a relative convergence that influences radiation characteristics, thereby indicating the existence of an optimal ratio for achieving quasiisotropic radiation while minimizing the mean absolute error (MAE).

Graph structure learning (GSL) aims to assist graph neural networks to yield effective node embeddings for downstream tasks, especially in scenarios with the absence of structures or the existence of unreliable edges. Most GSL models are built on the i.i.d. assumption across training and testing data. However, this assumption can be violated where test data contains out-of-distribution samples. Consequently, those models are limited in generalization, which will lead to a poor structure. On the other hand, while they have made great progress, additional optimized parameters are required due to their implementation with parametric models. To tackle the above problems, we propose a novel Generalizable and Non-parametric Structure learning framework named GNS, which can be easily and effectively applied to various tasks. GNS neither relies on i.i.d. assumption nor even involves any parameters being optimized, instead to find an appropriate similarity between nodes and an associated threshold to establish desirable structures. Specifically, we first incorporate the candidate neighbor distributions for nodes to refine the similarity. Then, we introduce an adaptive threshold discovery method inspired by Fisher's criterion to determine final structures. Extensive experiments demonstrate that GNS excels not only in OOD scenarios but also in the general classification and regression prediction tasks.