The use of SCADA and AMI systems in smart grid-based Industrial Internet-of-Things (SG-IIoT) networks for proper energy supply are noteworthy. Inaccurate energy load forecasts, cyber-threats, and energy load-based sustainability issues in smart grids hinder SG-IIoT operations. To mitigate these challenges, a federated-learning approach is developed by integrating LSTM (Long-Short-Term-Memory), Transformer-LLM (Larger Language Model) based Foundation-Model, and AES (Advanced-Encryption-Standard) cryptography. The proposed approach is named Responsible-Federated-Foundation-Model (ResFedFM). To ensure secure federated learning computation as well as data security at the edge (smart meter), fog (SCADA-based substation grid) and cloud (grid cloud server) layers of the SG-IIoT, a self-parent keys-based cryptography method has been developed by combining AES with HMAC (Hash-based-Message-Authentication-Code). A load forecasting algorithm called LSTM-LLM-GenResAI-Forecasting has been developed for computation at each end node of the federated learning process. The edge node forecast outputs are encrypted and aggregated at the fog node. At the fog node, the data are decrypted, and aggregation algorithm of federated-learning process are used to generate overall load forecasting of each sub-station grid. Again, the forecast data from these fog nodes are aggregated in an encrypted state at the cloud level and overall load forecasts are generated for multiple fog nodes. The result of proposed approach provides responsible forecasting (High accuracy, green computing-based energy demand, optimization of AI-hallucination, and grid data security), demonstrating enhanced performance over seven significant models.

Multi-carrier relay selection is of particular interest and challenge due to the spatio-frequency coupling and the dynamics of available spectral and relay resources. Among a number of promising relay selection schemes, combined relay selection stands out as an equilibrium between system complexity and reliability. Recent research progress has witnessed the capability of neural computing as a powerful tool to efficiently realize combined relay selection for a given network topology where the number of relays and their locations are fixed and known. However, for contemporary wireless networks that are highly dynamic, the classic neural computing methods can hardly help out because of the scale drifts of input and output matrices. To enable multi-carrier combined relay selection in stochastic wireless networks (SWNs) where the number of available relays for selection could vary, we propose a recurrent neural network (RNN) based framework and devise several training methods suited for various application scenarios. In addition, we conduct a set of computer experiments to verify the effectiveness and efficiency of the proposed RNN-based framework compared with several baselines. With the obtained experimental results, we also evaluate and discuss the proposed framework's reliability, generalization ability, and the robustness against imperfect channel state information (CSI).

The Internet of Medical Things (IoMT) mobile devices such as ambulance, medical done, and emergency mobile medical equipment face severe signal distortions due to interference, end-to-end packet loss, handoff delays, and lower throughputs during mobility. Network mobility basic support protocol (NBSP) has been proposed using the IP-based Wi-Fi solution to solve these issues. However, the weak signal, extra signaling overhead, and higherdelays were identified during handover due to patients' excessive requisites, resulting in radio link failure. Therefore, this article proposes a novel resource-efficient flow-enabled distributed mobility anchoring (FDMA) framework enhancing the functionalities of the centralized network entities and mobility entities.The performance of the proposed FDMA framework is evaluated and compared with the standard NBSP and proxy NEMO (PNEMO) scheme in terms of the variable number of cell residence time and mobile routers, where the proposed framework outperformed NBSP and PNEMO schemes for IoMT Mobile devices in 5G network.

Dynamic optimisation problems (DOPs) have attracted a lot of research attention in recent years due to their practical applications and complexity. DOPs are more challenging than static optimisation problems because the problem information or data is either revealed or changed during the course of an ongoing optimisation process. This requires an optimisation algorithm that should be able to monitor the movement of the optimal point and the changes in the landscape solutions. In this paper, we proposed an Interleaved Artificial Bee Colony (I-ABC) algorithm for DOPs. Artificial Bee Colony (ABC) is a nature inspired algorithm which has been successfully used in various optimisation problems. The proposed I-ABC algorithm has two populations, called ABC1 and ABC2, which worked in an interleaved manner. While ABC1 focused on exploring the search space though using a probabilistic solution acceptance mechanism, ABC2 worked inside ABC1 and focused on the search around the current best solutions by using a greedy mechanism. The proposed algorithm was tested on the Moving Peak Benchmark. The experimental results indicated that the proposed algorithm achieved better results than the compared methods for 8 out of 11 scenarios.

Recently, interest in solving real-world problems that change over the time, so called dynamic optimisation problems (DOPs), has grown due to their practical applications. A DOP requires an optimisation algorithm that can dynamically adapt to changes and several methodologies have been integrated with population-based algorithms to address these problems. Multi-population algorithms have been widely used, but it is hard to determine the number of populations to be used for a given problem. This paper proposes an adaptive multi-population artificial bee colony (ABC) algorithm for DOPs. ABC is a simple, yet efficient, nature inspired algorithm for addressing numerical optimisation, which has been successfully used for tackling other optimisation problems. The proposed ABC algorithm has the following features. Firstly it uses multi-populations to cope with dynamic changes, and a clearing scheme to maintain the diversity and enhance the exploration process. Secondly, the number of sub-populations changes over time, to adapt to changes in the search space. The moving peaks benchmark DOP is used to verify the performance of the proposed ABC. Experimental results show that the proposed ABC is superior to the ABC on all tested instances. Compared to state of the art methodologies, our proposed ABC algorithm produces very good results.

Electromagnetic algorithm is a population based meta-heuristic which imitates the attraction and repulsion of sample points. In this paper, we propose an electromagnetic algorithm to simultaneously tune the structure and parameter of the feed forward neural network. Each solution in the electromagnetic algorithm contains both the design structure and the parameters values of the neural network. This solution later will be used by the neural network to represents its configuration. The classification accuracy returned by the neural network represents the quality of the solution. The performance of the proposed method is verified by using the well-known classification benchmarks and compared against the latest methodologies in the literature. Empirical results demonstrate that the proposed algorithm is able to obtain competitive results, when compared to the best-known results in the literature.

Clustering is an important data analysis and data mining tool that is used in many fields and applications, which aims to find a homogeneous sets of objects based on the degree of similarity and dissimilarity of their attributes. One of the most popular techniques in data clustering is K-means, which is a simple, fast and efficient method that has been applied successfully in many fields. However, K-means has its own drawbacks like highly dependence on the initial solution and can easily trapped into local optima. In this paper, we investigate the behaviour of the newly created meta-heuristic optimisation algorithm called Biogeography-Based Optimisation (BBO) for data clustering with different initial solution generation mechanisms (random initial solution, sequential diversification initial solution, heuristic initial solution) that is based on the idea of migration of species between different habitats. To evaluate the performance of the proposed method, six UCI Machine Learning Repository data sets were used. The performance of the BBO algorithm was compared with well-known data-clustering algorithms that available in the literature, the experimental results showed that the BBO algorithm was able to obtain comparable results.

This paper presents a multi-objective post enrolment course timetabling problem as a new case study. We added a new soft constraint to the original single objective problem to both increase the complexity and represent a real world course timetabling problem. The new soft constraint introduced here attempts to minimize the total number of waiting timeslots in between courses for every student in a day. We proposed a Non-dominated Sorting Genetic Algorithm-II with a variable population size, called NSGA-II VPS, based on a given lifetime for each individual that is evaluated at the time of its birth. The algorithm was tested on the standard benchmark problems and experimental results show that the proposed method demonstrably improved upon the original approach (NSGA-II).

Hyper-heuristics can be defined as search method for selecting or generating heuristics to solve difficult problem. A high level heuristic therefore operate on a set of low level heuristics with the overall aim of selecting the most suitable set of low level heuristics at a particular point in generating an overall solution. In this work, we propose a set of constructive hyper-heuristics for solving attribute reduction problems. At the high level, the hyper-heuristics (at each iteration) adaptively select the most suitable low level heuristics using roulette wheel selection mechanism. Whilst, at the underlying low level, four low level heuristics are used to gradually, and indirectly construct the solution. The proposed hyper-heuristics has been evaluated on a widely used UCI datasets. Results show that our hyper-heuristic produces good quality solutions when compared against other metaheuristic and outperforms other approaches on some benchmark instances.

Attribute reduction is a basic issue in knowledge representation and data mining. It simplifies an information system by discarding some redundant attributes. In this paper, we present a hybrid approach that combines the nature of variable neighbourhood search in the first phase with an iterated local search in the second phase that always accepts best solutions. The approach is tested over 13 well-known established datasets. The results demonstrate that the variable neighbourhood search approach is able to produce solutions that are competitive with those state-of-the-art techniques from the literature in terms of minimal reducts.

In this paper we propose a rough classification modeling algorithm based on Ant Colony Optimization (ACO) reduction. We used ACO to compute the rough set reduct and later a modified rules generation method is employed to generate the classification rules. The rules generation algorithm used is the simplification of the Default Rules Generation Framework (DRGF) in order to fit with the ACO reduct. The performance of the proposed classifier is compared with the DRGF based classifier using genetic reduction. The experimental results show that the ACO-Rough performs better with higher classification accuracy and fewer number of rules.

Attribute reduction is one of the main issues in the theoretical research of rough set theory which is known as a NP-hard optimization problem. The objective is to find the minimal number of attributes from a large dataset. Hence it is difficult to solve to optimality. This paper proposes a composite neighbourhood structure approach to solve the attribute reduction problem that consists of two versions. The first version is a basic composite neighbourhood structure (CNS) approach where the neighbourhood is selected at random. For the second version, the selection of the neighbourhood structure is based on certain rules (coded as IS-CNS). Both of the algorithms only accept an improved solution. The proposed approach is tested on a set of benchmark datasets taken from University of California, Irvine (UCI) machine learning respiratory in comparison with a set of state-of-the-art methods from the literature. The experimental results show that the proposed approach is able to produce competitive results for the test datasets.

Combinations of evolutionary based approaches with local search have provided very good results for a variety of scheduling problems. This paper describes the development of such an algorithm for university course timetabling. This problem is concerned with the assignment of lectures to specific timeslots and rooms. For a solution to be feasible, a number of hard constraints must be satisfied. The quality of the solution is measured in terms of a penalty value which represents the degree to which various soft constraints are satisfied. This hybrid evolutionary approach is tested over established datasets and compared against state-of-the-art techniques from the literature. The results obtained confirm that the approach is able to produce solutions to the course timetabling problem which exhibit some of the lowest penalty values in the literature on these benchmark problems. It is therefore concluded that the hybrid evolutionary approach represents a particularly effective methodology for producing high quality solutions to the university course timetabling problem.