ChatGPT has transformed the way how education is accessed and made education more interactive and personalised; however, depending on the use, it can be effective or disruptive in education. The use of ChatGPT in education has brought both opportunities and challenges to users and academic institutes. It can make virtually all educational services personalised, automated, efficient and cost-effective; however, users and academic institutes should also be concerned about it using in education due to a number of issues with it, such as misinformation, inaccuracy, bias, cheating, privacy and security; therefore, a careful consideration should be given by them before using it in education. This paper will provide both effective and disruptive aspects of ChatGPT, and elucidate how it can be effective and how it can be disruptive in education. Initially, it will discuss the historical development of chatbots from ELIZA to ChatGPT, and applications of various AI chatbots in education. Afterwards, this paper will elucidate various advantages of ChatGPT in education. Lastly, it will elucidate various challenges of ChatGPT in education. This elucidation will offer a comprehensive and simplified analysis of educational use of ChatGPT.

State-of-the-art (SOTA) models typically use large datasets for pre-training and then fine-tune on smaller datasets for better performance. However, the high computational cost can be a barrier for many researchers. There is a need to focus on data size-independent models suited for data-scarce scenarios, which is essential for tasks like fingerprint recognition and could make research more accessible and generalizable in resource-limited environments. With this aim, this paper presents a novel approach to the difficulties in contactless fingerprint recognition, particularly with scarce and poor-quality challenging dataset images due to contactless acquisition. Our proposed system uses a 'Scattering using a Shearlet Network (SSNet)' to extract fingerprint features and a score-level fusion scheme to improve authentication accuracy. In contrast to the computationally expensive and mathematically less transparent dense deep learning networks such as vision transformers, attention networks, deep learning-based hybrid approaches, etc., SSNet is an economical framework with fixed filters. The SSNet is a replacement to the Scattering Wavelet Network (SWN) that utilizes a Complex Morlet Wavelet (CMW). Our model significantly improves verification and identification accuracy over SOTA approaches, particularly with scarce and poor-quality challenging datasets.

Background: Relaxation methods determine residual stresses by measuring the deformations produced by incremental removal of a subdomain of the specimen. Measured strains at any given increment, determined by the cumulative effect of the relieved stresses, appear as an integral equation, which must be inverted to obtain residual stresses. In practice, stress distributions are discretized by a finite-dimensional basis, to transform the integral equations into a linear system of equations, which is often ill-conditioned. Objective: This article demonstrates that the problem is actually ill-posed and comes with an inherent bias-variance tradeoff. Methods: The hole drilling method is used as an example application, and the practical effects of ill-posedness are illustrated. Results: Traditional regularization of the solution by limiting the resolution of the discretization reduces solution variance (noise) at the expense of increased bias and often results in the ultimately harmful practice of taking fewer data points. A careful analysis including the alternate Tikhonov regularization approach shows that the highest number of measurements should always be taken to reduce the variance for a given regularization scheme. Unfortunately, the variability of a regularized solution cannot be used to build a valid confidence interval, since an unknown bias term is always present in the true overall error. Conclusions: The mathematical theory of ill-posed problems provides tools to manage the bias-variance tradeoff on a reasonable statistical basis, especially when the statistical properties of measurement errors are known. In the long run, physical arguments that provide constraints on the true solution would be of utmost importance, as they could regularize the problem without introducing an otherwise unknown bias. Constraining the minimum length scale to some physically meaningful value is one promising possibility.

Machine learning approach to estimate a probability distribution of trajectories in a dataset.

Deep Neural Network (DNN) belongs to an important class of machine learning algorithms generally used to classify digital data in the form of image and speech recognition. The computational complexity of a DNN-based image classifier is higher than traditional fully connected (FC) feed-forward NNs. Therefore, dedicated cloud servers and Graphical Processor Units (GPU) are utilized to achieve high-speed and large-capacity computation tasks in machine vision systems. However, a growing demand exists for real-time processing of complex machine-learning tasks on embedded systems. As FC layers consume the highest fraction of computational power and memory footprint, innovating novel power-efficient and low-footprint NN architecture for embedded systems is crucial. A novel design strategy and algorithms are proposed in this article, where a power-efficient FC DNN is implemented using a pipelined and parallel Fast Fourier Transform (FFT) on a circular projection-based architecture. The footprint of the DNN is further reduced using a folded FFT network. The proposed algorithm is tested using two benchmark training set examples, the “MNIST database of handwritten digits” and the “CIFAR-10 database”. In both cases, we achieved > 90% accuracy, while the power consumption of the network is 37% less than the traditional FFT architecture-based DNNs.

In modern Electrical Power Distribution System (EPDS), power electronics devices are gaining more prominence. Three-phase ac/dc converters provide a fully-controlled interface between the onboard ac and dc electrical equipment. In the context of utility ac electrical power systems, impedance-based stability analysis under Synchronous Reference Frame (SRF), or dq impedance has been established as a powerful tool to evaluate stability and optimize control design. However, precise measurement of the dq impedance is challenging in practice especially for high-capacity or high ac fundamental frequency systems. In this paper, a hardware-in-the-loop (HiL) platform is proposed for measuring dq impedance. An aircraft ac EPDS bus is built with a grid forming (GFM) converter and an active front end (AFE) converter to verify the proposed scheme. A comparison is made between the HiL measurement results, the experimental results and the analytical model, validating that dq-impedance measurement with the proposed setup provides a precise and efficient solution that obviates the need for complete hardware experiments.

Performance modelling of a deep learning application is essential to improve and quantify the efficiency of the model framework. However, existing performance models are mostly case-specific, with limited capability for the new deep learning frameworks/applications. In this paper, we propose a generic performance model of an application in a distributed environment with a generic expression of the application execution time that considers the influence of both intrinsic factors/operations (e.g. algorithmic parameters/internal operations) and extrinsic scaling factors (e.g. the number of processors, data chunks and batch size). We formulate it as a global optimization problem and solve it using regularization on a cost function and differential evolution algorithm to find the best-fit values of the constants in the generic expression to match the experimentally determined computation time. We have evaluated the proposed model on three deep learning frameworks (i.e., TensorFlow, MXnet, and ytorch). The experimental results show that the proposed model can provide accurate performance predictions and interpretability. In addition, the proposed work can be applied to any distributed deep neural network without instrumenting the code and provides insight into the factors affecting performance and scalability.

As the number of distributed energy resources in today’s energy distribution grids is growing, measures must be taken to keep the voltage magnitudes in the grid within the operational limits. One prevalent way of doing this is to use voltage dependent reactive power provisioning, so-called Q(V )-control, where a Q(V )-characteristic with a certain slope prescribes how much reactive power is used to stabilise the node voltage. However, due to their interconnection by the power grid, the combined Q(V )-controls of several plants may become unstable. The first part of this paper revisits an existing stability criterion from the literature, based on a discrete-time model of Q(V)-control but limited to radial topologies, in order to assess how steep the slopes can be chosen. In the second part, this criterion is extended to grids with arbitrary topologies by introducing a more general grid model which explicitly takes the current operating point of the grid into account. Furthermore, an optimisation-based procedure is presented to exploit this new criterion for the direct maximisation of the slopes of the Q(V)-characteristics while guaranteeing stability of the system. Finally, both criteria are compared by applying them on selected benchmark grids.

Routing Protocol for Low-power and lossy networks (RPL), as the de-facto routing protocol for IoT networks, neglects to exploit IoT devices’ full capacity to tune their transmission power. One of the reasons is that optimizing the transmission power in parallel with the routing strategy is challenging, given the dynamic nature of wireless links and the constrained resources in IoT devices. Optimizing the transmission power requires evaluating the probability of packet collisions, energy consumption, the number of hops, and interference. We propose Adaptive Control of Transmission pOwer for RPL (ACTOR) for dynamic optimization of transmission power. ACTOR aims at improving throughput in dense networks by passively exploring different transmission power levels. The extent of resources used for this exploration significantly affects the network throughput. Thus, the exploration needs to adapt to dynamism in the environment. We formulate this exploration strategy using the Multi-Armed Bandit framework. The classic solutions of bandit theory including Upper Confidence Bound and Discounted Upper Confidence Bound accelerate the convergence of the exploration and guarantee its optimality. We also enhance ACTOR by mechanisms from RPL to blacklist undesirable transmission power levels and stabilize the topology. Results of the experiments on our 40-node testbed and simulations show that ACTOR achieves higher throughput (increasing the packet delivery ratio by 20%), energy consumption, end-to-end delay, and the number of retransmissions are significantly improved against the standard RPL and the selected benchmark.

A document by Chaoyu Zhang. Click on the document to view its contents.

While deep learning and backpropagation continue to dominate the field of machine learning in terms of benchmarks and versatility, recent neuroscientific advances shed light on more biologically plausible approaches. Spiking neural networks (SNNs), modelled after action potential dynamics, offer inherent time sensitivity and more efficiency in terms of performance to complexity. While investigating paradigms to support such alternatives, we attempt to answer whether reservoir computing can benefit from a spiking network based implementation with elements of biologically realistic models. This is done by varying both hyperparameters and reservoir generation approaches and comparing implementations to spot potential improvements. We demonstrate how customized training of SNNs can result in competitive performance levels at lower operational complexity and be readily applied to other paradigms, such as the development of reservoir dynamics.

The problem of computing machine passing the maze is one of theoretical computer science key tasks. This task was partially considered by the classics of computer science, e.g. A. Turing, C. Shannon, L. Budach and Z. Pawlak. Labyrinth problem solving includes fundamental knowledges in the main three branches, namely  environment (labyrinth) knowledge, computing machines features knowledge and behavior of computing machines in labyrinth knowledge. These three components are built on the fundamental basis of the theoretical computer science. Nowadays the labyrinth passing task is complemented by intellectuality as an additional criterion. That is, computing machines must acquire intelligent functions, and this task gets a new component and a new statement, which is briefly formulated as the search for a way out of the labyrinth by intelligent computing machines. The main concept of this article is to develop artificial neural networks based intelligent functions for calculating machine to pass the maze. Besides this visual aspect is emphasized, i.e. the computational machine includes a maze viewing function. Authors accentuate fundamental principles of artificial neural networks technologies building  and engage backpropagation algorithm, which is used in a artificial neural network learning process. The article discusses maze, dataset construction, artificial neural network training, maze recognition computational experiment and analysis of the hyperparameters effect on artificial neural network training. The aim of the article is to build mathematical model based on backpropagation algorithm and to identify how hyperparameters affect the learning process of artificial neural networks. Thus the research is based on the methods of computational experiment and step-by-step detailing to implement the algorithm. The results of the research include theoretical basis of computing machines intellectualization for the maze and computing machine ability to recognize the maze. These results may be useful to theorists for a detailed description of the process and to practitioners for ability to test this algorithm and approach while solving a problem of maze passing by intelligent agents.

Accurate state-of-health (SOH) estimation is crucial for the reliable operation of lithium-ion batteries in sustainable energy systems. While existing methods often require complex feature engineering or focus on specific voltage ranges, this paper presents a novel and practical approach using readily available discharge data from Battery Management Systems (BMS). We introduce a hybrid feature set that combines the strengths of incremental capacity (IC) curves and temperature generation factors, capturing critical aspects of battery degradation. By employing a two-layer feedforward Artificial Neural Network (ANN) and analyzing the complete discharge cycle, our method was tested two different dataset, achieves an average root mean square error (RMSE) of 0.38% across two diverse battery degradation datasets with minimal computational cost. This approach offers a promising solution for real-world on-board SOH estimation, enabling efficient battery management and ensuring the longevity of energy storage systems.

We investigate the design of quantized event-triggered controllers for LTI systems with unknown initial states. The proposed approach is novel and incorporates interesting features. First, the quantizer that we synthesize is dynamic with exponential behaviour, which enables us to capture the state in a finite time. Second, once the state is captured, the quantization error is guaranteed to remain bounded thanks to the sliding mode feature. Third, the quantizer dynamics only depend on the sign of the quantization error, which allows for one-bit transmission rather than packet-based transmission. Moreover, the event-triggering mechanism that we consider is also dynamic and depends only on locally available information. The overall system is modeled as a hybrid dynamical system and the closedloop stability is investigated using Lyapunov functions. The approach ensures global asymptotic stability property for the closed-loop system. Moreover, Zeno behavior is prevented by means of time regularization techniques. The effectiveness of the proposed approach is illustrated by numerical simulation.

As human-robot interaction rapidly spreads in numerous fields, the subject of robot acceptance gains increasing importance. Visual similarity to the human body, as occurs for humanoids, is generally not enough to guarantee acceptance in physical interaction, as acceptance directly links to comfort and ergonomics, which are measured in terms of the quality of the robot movement perceived by the human. This paper discusses the connection between comfort and similarity of the robot movement to the human one. By considering the kinematic characterization of human movement, this paper focuses on the time laws of such movements, wherein the end-effector path is prescribed. Based on the sigma-lognormal velocity model, a human-likeness index is defined and used to provide with an a priori characterization of trajectories. Such an index can be used to evaluate the performance of trajectory generation algorithms in producing human-like movements, before they are actually executed. For validation purposes, through physical interaction with a robot, a sample of 38 subjects is asked to compare trajectories and judge about their comfort over three experimental campaigns. The results demonstrate a globally consistent trend between the preference in terms of perceived comfort and the distribution of the suggested human-likeness index.

The eNeuron project is an ongoing Horizon 2020 EU-funded project with the main objective of setting out a practical framework for optimising the design and operation of local energy communities (LECs), acting as multi-carrier energy hubs (EH) and micro-energy hubs (mEH). LECs are key in the transition towards green, decarbonised power from local and renewable sources. The consortium is developing and applying pioneering software and hardware solutions, putting them to the test in real energy communities and introducing new governance models. There are four demonstration Pilots in the eNeuron Project: The Polish City of Bydgoszcz, the Skagerak Energilab in Norway, the Lisbon Naval Base (LNB) in Portugal, and the UNIVPM University Campus in Italy. This paper focuses on presenting the Portuguese demo pilot: the Lisbon Naval Base, namely the development of the demonstration activities so far, roughly half of the duration of the project, as well as a balance of all supporting activities, such as Use Case definition, technical activities (asset definition and installation) and social activities (stakeholder engagement). This paper presents a description and conclusion of these activities so far, at the time that the eNeuron preliminary toolbox concept is ready to be shared and implemented in the Pilots.

Despite the emergence of numerous Deep Learning (DL) models for breast cancer detection via mammograms, there is a lack of evidence about their robustness to perform well on new unseen mammograms. To fill this gap, we introduce StethoNet, a DL-based framework that consists of multiple Convolutional Neural Network (CNN) trained models for classifying benign and malignant tumors. StethoNet was trained on the Chinese Mammography Database (CMMD), and tested on unseen images from CMMD, as well as on images from two independent datasets, i.e., the Vindr-Mammo and the INbreast datasets. To mitigate domain-shift effects, we applied an effective entropy-based domain adaptation technique at the preprocessing stage. Furthermore, a Bayesian hyperparameters optimization scheme was implemented for StethoNet optimization. To ensure interpretable results that corroborate with prior clinical knowledge, attention maps generated using Gradient-weighted Class Activation Mapping (GRADCAM) were compared with Regions of Interest (ROIs) identified by radiologists. StethoNet achieved impressive Area Under the receiver operating characteristics Curve (AUC) scores: 90.7% (88.6%-92.8%), 83.9% (76.0%-91.8%), and 85.7% (82.1%-89.4%) for the CMMD, INbreast, and Vindr-Mammo datasets, respectively. These results surpass the current state of the art and highlight the robustness and generalizability of StethoNet, scaffolding the integration of DL models into breast cancer mammography screening workflows.

The transition from 5G to 6G will result in an explosion of Internet of Things (IoT) devices that provide pervasive and constant connection across all spheres of human activities. In this regard, non-terrestrial networks (NTNs) will play a crucial role in supporting and supplementing terrestrial systems in order to cope with such a vast number of IoT devices and to fulfill the massive capacity needs of the most sophisticated of them. This study investigates the implication of several multiple access techniques for 6G enabled NTN-assisted IoT technologies. First, general architecture of NTN, NTN-assisted IoT technologies, and key features and challenges of NTNs are presented. Then, different types of multiple access schemes are discussed and compared in the context of NTN-assisted IoT systems. Simulation results are presented and important performance parameters such as energy efficiency and spectrum efficiency are examined for the discussed multiple access schemes. Furthermore, challenges and future research directions are discussed.