Cloud computing (CC) delivers processing power and data storage on demand. It is one of the most significant computer science technologies, contributing to healthcare, industry, and the Internet of Things. One of CC's biggest security concerns are intrusion detection and separating harmful from legitimate communication, similar to computer networks. Although a wide range of intrusion detection systems is available today, they often suffer from misclassification issues, where the system can fail to recognize an attack as a threat or to mark normal traffic as malicious. This research proposes classifying network traffic using a convolutional neural network and extreme gradient boosting model. Additionally, a modified sine cosine algorithm is used to tune model hyperparameters for optimal performance. The presented framework was tested on major real-world TON IoT intrusion detection datasets. The proposed optimizer is compared to many recent metaheuristics in a matched experimental setting. The simulation results show that the suggested technique is superior to other methods for both datasets, with the best-performing optimized models achieving an accuracy of 96.667 on Windows 10 and 98.6731 on Windows 7 simulation.

Neurodegenerative conditions are defined by the progressive deterioration and death of nerve cells in the core neural system. Most neurodegenerative conditions are not curable. While there have been significant improvements and techniques used to treat these diseases early diagnosis continues to play a crucial role in the entire approach. Conditions are often diagnosed only once they start negatively impacting the daily life of those affected. Early detection and timely preventative treatment can help improve patient subjective well-being. This study examines the application of a non-invasive gait analysis technique for the detection of Parkinson’s disease. Publicly available data collected from patients suffering from Parkinson’s along with control groups is utilized and combined with long-short-term neural networks to construct models capable of detecting signs on Parkinson’s disorder. However, because of the significant reliance of models on appropriate parameters selection, metaheuristic algorithms are used to fine tune the selection process, and a modified variation of the strongly founded PSO algorithm was proposed. Several contemporary optimizers are compared based on their ability to optimize model performance. This suggested approach achieved the superior outcomes with an accuracy of 89.92%. The constructed models have been evaluated to determine feature importance using game theory based methods.

Abstract The progress of Industrial Revolution 4.0 has been supported by recent advances in several domains, and one of the main contributors is the Internet of Things. Smart factories and healthcare have both benefited in terms of leveraged quality of service and productivity rate. However, there is always a trade‐off and some of the largest concerns include security, intrusion, and failure detection, due to high dependence on the Internet of Things devices. To overcome these and other challenges, artificial intelligence, especially machine learning algorithms, are employed for fault prediction, intrusion detection, computer‐aided diagnostics, and so forth. However, efficiency of machine learning models heavily depend on feature selection, predetermined values of hyper‐parameters and training to deliver a desired result. This paper proposes a swarm intelligence‐based approach to tune the machine learning models. A novel version of the firefly algorithm, that overcomes known deficiencies of original method by employing diversification‐based mechanism, has been proposed and applied to both feature selection and hyper‐parameter optimization of two machine learning models—XGBoost and extreme learning machine. The proposed approach has been tested on four real‐world Industry 4.0 data sets, namely distributed transformer monitoring, elderly fall prediction, BoT‐IoT, and UNSW‐NB 15. Achieved results have been compared to the results of eight other cutting‐edge metaheuristics, that have been implemented and tested under the same conditions. The experimental outcomes strongly indicate that the proposed approach significantly outperformed all other competitor metaheuristics in terms of convergence speed and results' quality measured with standard metrics—accuracy, precision, recall, and f1‐score.

Bitcoin price volatility fascinates both researchers and investors, studying features that influence its movement. This paper expends on previous research and examines time series data of various exogenous and endogenous factors: Bitcoin, Ethereum, S&P 500, and VIX closing prices; exchange rates of the Euro and GPB to USD; and the number of Bitcoin-related tweets per day. A period of three years (from September 2019 to September 2022) is covered by the research dataset. A two-layer framework is introduced tasked with accurately forecasting Bitcoin price. In the first layer, to account for complexities in the analyzed data, variational mode decomposition (VMD) extracts trends from the time series. In the second layer, Long short-term memory and hybrid Bidirectional long short-term memory networks were used to forecast prices several steps ahead. This work also introduced an enhanced variant of the sine cosine algorithm to tune the control parameters of VMD and both neural networks for attaining the best possible performance. The main focus is on combining VMD with modified metaheuristics to improve cryptocurrency closing value forecast. Two sets of experiments were conducted, with and without VMD. The results have been contrasted with models tuned by seven other cutting-edge optimizers. Extensive experimental outcomes indicate that Bitcoin price can be forecasted with great accuracy using selected features and time series decomposition. Additionally, the best model was analyzed, and Shapley values indicated that features such as EUR/USD exchange rates, Ethereum closing prices, and GBP/USD exchange rates, have a significant impact on forecasts.

Abstract Recently, the Honey Badger Algorithm (HBA) was proposed as a metaheuristic algorithm. Honey badger hunting behaviour inspired the development of this algorithm. In the exploitation phase, HBA performs poorly and stagnates at the local best solution. On the other hand, the sand cat swarm optimization (SCSO) is a very competitive algorithm compared to other common metaheuristic algorithms since it has outstanding performance in the exploitation phase. Hence, the purpose of this paper is to hybridize HBA with SCSO so that the SCSO can overcome deficiencies of the HBA to improve the quality of the solution. The SCSO can effectively exploit optimal solutions. For the research conducted in this paper, a hybrid metaheuristic algorithm called HBASCSO was developed. The proposed approach was evaluated against challenging CEC benchmark instances taken from CEC2015, CEC2017, and CEC2019 benchmark suites The HBASCSO is also evaluated concerning the original HBA, SCSO, as well as several other recently proposed algorithms. To demonstrate that the proposed method performs significantly better than other competitive algorithms, 30 independent runs of each algorithm were evaluated to determine the best, worst, mean, and standard deviation of fitness functions. In addition, the Wilcoxon rank-sum test is used as a non-parametric comparison, and it has been found that the proposed algorithm outperforms other algorithms. Hence, the HBASCSO achieves an optimum solution that is better than the original algorithms.

Abstract This study explores crop yield forecasting through weight agnostic neural networks (WANN) optimized by a modified metaheuristic. WANNs offer the potential for lighter networks with shared weights, utilizing a two-layer cooperative framework to optimize network architecture and shared weights. The proposed metaheuristic is tested on real-world crop datasets and benchmarked against state-of-the-art algorithms using standard regression metrics. While not claiming WANN as the definitive solution, the model demonstrates significant potential in crop forecasting with lightweight architectures. The optimized WANN models achieve a mean absolute error (MAE) of 0.017698 and an R-squared ( $$R^2$$ R 2 ) score of 0.886555, indicating promising forecasting performance. Statistical analysis and Simulator for Autonomy and Generality Evaluation (SAGE) validate the improvement significance and feature importance of the proposed approach.

Energy forecasting plays an important role in effective power grid management. The widespread adoption of emerging technologies and the increased reliance on renewable sources of energy have created a need for a robust and accurate system for energy forecasting. This demand is becoming increasingly relevant due to the ongoing 2022 energy crisis. Modern power systems are very complex with many complicated correlations between various forecasting factors and parameters. Furthermore, renewable energy is often dependent on weather conditions, which complicates the process of forecasting. This work presents a novel artificial intelligence (AI) driven energy forecasting tuned deep learning framework. By formatting predictors as a time series, two variations of recurrent neural networks (RNN)s have been implemented: long-short-term memory (LSTM) and gated recurrent unit (GRU) neural networks. However, both approaches present several hyperparameters that require adequate tuning to attain desirable performance. Therefore, this work also proposes an improved version of a well know swarm intelligence algorithm, the sine cosine algorithm (SCA), tasked with tackling hyperparameter tuning for both approaches. To demonstrate the improvements made, three datasets have been constructed for evaluation from publicly available real-world data that contain relevant solar, wind, and power-grid load parameters alongside weather data. The proposed metaheuristic algorithm has been subjected to a comparative analysis with several contemporary metaheuristic algorithms to showcase the improvements made. The introduced metaheuristic demonstrated the best performance with a mean square error (MSE) rate for solar generation of only 0.0132 with LSTM methods and 0.0134 with GRU. Similar performance was observed for wind power generation forecasting with a MSE of 0.00292 with LSTM and 0.00287. When tackling power grid load forecasting a median MSE of 0.0162 was attained with LSTM and 0.01504 with GRU. Therefore there is great potential for tackling these tasks using the proposed approach. The best-performing models have been analyzed using SHapley Additive exPlanations (SHAP) to determine the factors that have the highest influence on energy generation and demand.

In the last few decades, the World Wide Web has become a necessity that offers numerous services to end users. The number of online transactions increases daily, as well as that of malicious actors. Machine learning plays a vital role in the majority of modern solutions. To further improve Web security, this paper proposes a hybrid approach based on the eXtreme Gradient Boosting (XGBoost) machine learning model optimized by an improved version of the well-known metaheuristics algorithm. In this research, the improved firefly algorithm is employed in the two-tier framework, which was also developed as part of the research, to perform both the feature selection and adjustment of the XGBoost hyper-parameters. The performance of the introduced hybrid model is evaluated against three instances of well-known publicly available phishing website datasets. The performance of novel introduced algorithms is additionally compared against cutting-edge metaheuristics that are utilized in the same framework. The first two datasets were provided by Mendeley Data, while the third was acquired from the University of California, Irvine machine learning repository. Additionally, the best performing models have been subjected to SHapley Additive exPlanations (SHAP) analysis to determine the impact of each feature on model decisions. The obtained results suggest that the proposed hybrid solution achieves a superior performance level in comparison to other approaches, and that it represents a perspective solution in the domain of web security.

The assessment of the degradation state of an unearthed ancient artefact concerns the identification of the material composition and of the corrosive compounds that are present at the surface. The standard investigation makes use of a combination of invasive and non-invasive techniques and complex devices, while it also relies on the extensive experience of the restorer. The current paper puts forward a new possibility of employing an alternative computational solution with the support of deep learning that recognizes and delineates all the corrosive compounds from stereo microscope images of the surface of a metal item. With the input received from a portable microscope, such a fast non-destructive tool would provide straightforward assistance at the excavation site, as well as a second opinion for novice investigators. Iron archaeological objects are considered and four corrosion compounds are identified and outlined by the deep learning models, i.e. Fe2O3, FeSO4, FeCl3 and FeO. The results show that the deep computational identification and delimitation of the four corrosive types is meticulous even with a minimal annotation provided for training.

Fetal echocardiography has become a recommended investigation during pregnancy routine examination in the second semester of fetal development. The heart structures are now sufficiently developed to detect any indication of a possible congenital heart disease. However, the inspection of the key components is not as straightforward for the less experienced young obstetricians. Moreover, a background automatic check of the presence of the main elements could be helpful in speeding up the overall assessment during the periodic pregnancy control. In this respect, the current paper proposes the support of a deep learning model towards the semantic segmentation of eleven key structures that need to be identified within the second trimester fetal heart scans.

In the area where privacy is of greater concern, federated learning,a distributed machine learning strategy for preserving privacy,is widely employed in several privacy concern applications. In the meantime, neural architectures became familiar with deep learning approaches for automatic tuning of the architecture of deep neural networks (DNN). While searching with neural architecture and federated learning has experienced several challenges, optimized neural architecture research in federated learning is extensively on demand. DNN faces numerous issues while training such user privacy and ensuring the integrity of the aggregated results obtained from a server. To provide solutions for the above-mentioned issues, enormous federated learning techniques worked towards preserving privacy and were applied in different situations. Still, it is an open challenge that enables users to verify if the cloud server functions appropriately while ensuring users’ privacy while training. Federated Learning Method is a new way to improve the accuracy and precision, since the previous approach failed to opt the solutions. Here, Elliptical Curve Cryptography with Blockchain-based Federated Learning (ECC-BFL)is proposed to ensure the confidentiality of users’ local gradients while performing federated learning. The parameters such as classification accuracy, running time, Communication overhead, Computation overhead, and transaction speed are considered. The values obtained for these parameters are compared against three standard methods, namely Biparing Method (BM) Homomorphic Cryptosystem (HC), and Multiple Authorities with Attribute-Based Signature scheme (MA-ABS)against proposed Elliptical Curve Cryptography with Blockchain-based Federated Learning (ECC-BFL). As a result, the proposed ECC-BFL achieved 95% of classification accuracy, 65 sec of running time, 76% of communication overhead, 63% of computation overhead, and 92% of transaction speed.

In recent years, deep learning has reached exceptional accomplishment in diverse applications, such as visual and speech recognition, natural language processing. The convolutional neural network represents a particular type of neural network commonly used for the task of digital image classification. A common issue in deep neural network models is the high variance problem, or also called over-fitting. Over-fitting occurs when the model fits well with the training data and fails to generalize on new data. To prevent over-fitting, several regularization methods can be used; one such powerful method is the dropout regularization. To find the optimal value of the dropout rate is a very time-consuming process; hence, we propose a model to find the optimal value by utilizing a metaheuristic algorithm instead of a manual search. In this paper, we propose a hybridized bat algorithm to find the optimal dropout probability rate in a convolutional neural network and compare the results to similar techniques. The experimental results show that the proposed hybrid method overperforms other metaheuristic techniques.

We have recently witnessed the rapid development of several emerging technologies, including the internet of things, which lead to a high interest in wireless sensor networks. Tiny sensor nodes are now important parts of a large number of complex systems, with numerous applications including military, environment monitoring, surveillance and body area sensor networks. One of the biggest challenges each wireless sensor network has to handle is the network lifetime maximization. To achieve this, numerous clustering algorithms have been created, with the goal to improve energy consumption throughout the network by balancing the energy consumption overall nodes. All clustering algorithms incorporate load balancing to achieve energy efficiency. One of the basic and most important algorithms in use is LEACH. Swarm intelligence metaheuristics have already been applied in solving numerous problems of wireless sensor networks, including lifetime optimization, localization and many other NP hard problems with promising results, as can be seen in the literature overview. In the research proposed in this paper, an improved version of the firefly algorithm has been applied to improve the network lifetime. The firefly algorithm was used to help in forming the clusters and selection of the cluster head. Additionally, we have evaluated the performance of the improved firefly algorithm by comparing it to the LEACH, basic firefly algorithm and particle swarm optimization, that were all tested on the same network infrastructure model. Conducted simulations have proven that our proposed metaheuristic achieves better and more consistent performance than other algorithms.

The cloud computing is an emerging paradigm that enables dynamic provision of elastic, scalable and distributed computer resources to the end-users. One of the most important tasks of the cloud service provider is to deliver services to the end-users in an efficient manner from a finite pool of available physical and virtual resources. The efficiency in terms of both, cost-efficiency and computational-efficiency, can be accomplished through the load scheduling that has significant impact on the overall cloud system performance and represents one of the most important challenges in this domain. In this paper, we introduce two implementations of the original and improved versions of the tree growth algorithm for load scheduling in cloud computing environments. Tree growth algorithm is classified as swarm intelligence metaheuristic, that are able to successfully tackle NP hard problems such as cloud load scheduling. Both algorithms are implemented in the CloudSim environment and comparative analysis with other techniques and metaheuristics for this problem was performed. According to the obtained results, the improved version of the tree growth algorithm outperformed all other techniques and can be successfully applied to load scheduling in cloud systems.

Machine learning algorithms are used in various application and the need for faster and more accurate algorithms is urgent. Classification problem, as one of the most common machine learning tasks, has numerous proposed algorithms for solving it. One of the main factors that affects the classification accuracy, regardless of the used classifier, is the chosen feature set. Due to the fact that the classification quality depends on the features, feature selection represents an important task in machine learning. In this paper we propose adjusted bare bone fireworks algorithm for feature selection. Support vector machine is used as classifier, thus we additionally added SVM parameter optimization. The proposed method is tested on standard benchmark classification datasets from the UCI repository and the results are compared with other swarm intelligence methods. The results show that the proposed method achieves higher accuracy compared to the other methods even without SVM parameters tuning. As expected, parameter tuning has additionally increased the classification accuracy.

Digital images are part of numerous applications and they are used for various tasks. One of the important research topics in digital image processing is image fusion that aims to create more informative image by fusing two or more images from different sources or different points of view. The fused image is obtained as a linear combination of source images. Finding the optimal image fusion parameters, i.e. scaling factors, represents a hard optimization problem and in this paper we proposed brain storm optimization algorithm for solving it. Image fusion is done in frequency domain and the scaling factors for DCT coefficients of the source images are searched for. The proposed brain storm optimization method was tested with thermal and visual images. Standard benchmark dataset from TNO thermal image fusion was used for determining the quality of the proposed method. By comparing the obtained results it can be concluded that the proposed brain storm optimization method outperformed particle swarm optimization method, as well as several standard image fusion methods such as gradient pyramid, Laplacian pyramid, ratio of Laplacian pyramid and shift invariant discrete wavelet transform, in terms of four quality metrics including entropy, QABF, LABF and NABF.

This paper presents tree growth algorithm framework for designing convolutional neural network architecture. Convolutional neural networks are a special class of deep neural networks that typically consist of several convolution, pooling and fully connected layers. Convolutional neural networks have proved to be a robust method for tackling various image classification tasks. One of the most important challenges from this domain is to find the network architecture that has the best performance for the specific application. The performance of the network depends on the set of hyper-parameter values such as the number of convolutional and dense layers, the number of kernels per layer and kernel size. Optimization of hyperparameters was performed by novel tree growth algorithm that belongs to the group of swarm intelligence metaheuristics. The robustness, performance and solutions quality of the proposed framework was validated against the well-known MNIST dataset. Conducted comparative analysis demonstrated that the proposed frameworks obtains promising results in this domain.

Portfolio selection problem that deals with the optimal allocation of capital is a well-known hard optimization problem in the domains of economics and finance. Basic version of the problem is multi-objective since it deals with maximization of return with simultaneous minimization of risk. Additional real world constraints, including cardinality, make the problem even harder. Many techniques and heuristics have been applied to this intractable optimization problem, however swarm intelligence algorithms have been implemented only few times for this task, even though they are known to be very successful for that class of problems. In this paper, we hybridized artificial bee colony algorithm with elements inspired by genetic algorithms to obtain better balance between intensification and diversification, especially during late stages, and applied the proposed improved algorithm to the cardinality constrained mean-variance version of the portfolio selection problem. Experimental results on standard benchmark datasets from five stock indexes and comparative analysis with other cutting edge algorithms have shown that our proposed algorithm achieved better results considering all relevant metrics i.e. mean Euclidean distance between standard efficiency frontier and heuristic efficiency frontier from sets of Pareto optimal portfolios obtained by tested algorithms, mean return error and variance of return error.

In this paper we present bare bones fireworks algorithm implemented and adjusted for solving radio frequency identification (RFID) network planning problem. Bare bones fireworks algorithm is new and simplified version of the fireworks metaheuristic. This approach for the RFID network planning problem was not implemented before according to the literature survey. RFID network planning problem is a well known hard optimization problem and it poses one of the most fundamental challenges in the process of deployment of the RFID network. We tested bare bones fireworks algorithm on one problem model found in the literature and performed comparative analysis with approaches tested on the same problem formulation. We also performed additional set of experiments where the number of readers is considered as the algorithm's parameter. Results obtained from empirical tests prove the robustness and efficiency of the bare bones fireworks metaheuristic for tackling the RFID network planning problem and categorize this new version of the fireworks algorithm as state-of-the-art method for dealing with NP-hard tasks.

Robots have found their purpose in various situations, from speeding the manufacturing processes to performing complicated tasks in dangerous and hostile environments. One of the important problems in robotics is mobile robot path planning. Robot path planning represents a hard optimization problem that needs to be solved in numerous applications. In this paper we propose path planning method in environments with static obstacles based on the recent swarm intelligence algorithm, brain storm optimization. The brain storm optimization algorithm was improved by local search procedure that each new candidate solution moves to the local best position thus reducing computational time. We tested the proposed method on several benchmark examples from the literature and it has been shown that our approach finds better and more consistent paths using less computational time.

Wireless sensor networks are widely used and consequently represent an important research field. The objective of the node localization problem, that belongs to the group of NP-hard tasks, is to find geographical coordinates of each sensor node with unknown position that are randomly deployed in the monitoring area. Such hard optimization problems are successfully solved by the swarm intelligence algorithms. This paper presents hybridized recent swarm intelligence moth search algorithm adapted for solving localization problem in wireless sensor networks. The application of the moth search algorithm for node localization problem was not found in the literature survey. In the experimental section of this paper we show comparative analysis between the original and hybridized moth search algorithm, as well as with other state-of-the-art algorithms that were tested on the same problem instances of node localization problem. According to experimental results, both, basic moth search and hybridized moth search algorithms are promising approaches for dealing with this kind of problem.

In this work, target localization problem in adverse indoor environments is addressed, where most (if not all) links are non-line-of-sight (NLOS). Localization accuracy in such environments is highly affected by multipath, which makes the problem very challenging. Hence, in order to enhance the localization accuracy, received signal strength (RSS) and time of arrival (TOA) integrated measurements, are considered here. Nevertheless, the derived joint maximum likelihood (ML) problem is highly non-convex and has no closed-form solution; thus, some approximations are required to solve it. We show that, for small noise power, the ML estimator can be tightly approximated by another (non-convex in general) one, given in a form of a generalized trust region sub-problem (GTRS). Hence, exact solution of the derived estimator can be readily obtained by merely a bisection procedure. The proposed algorithm is compared with the state-of-the-art (SOA) RSS/TOA algorithms, as well as its RSS-only and TOA-only complements. Our simulations validate the effectiveness of the proposed approach, outperforming the SOA algorithms in all considered scenarios, and show the benefit of the measurement fusion.

Firefly algorithm is one of the recent and very promising swarm intelligence metaheuristics for tackling hard optimization problems. While firefly algorithm has been proven on various numerical and engineering optimization problems as a robust metaheuristic, it was not properly tested on a wide set of constrained benchmark functions. We performed testing of the original firefly algorithm on a set of standard 13 benchmark functions for constrained problems and it exhibited certain deficiencies, primarily insufficient exploration during early stage of the search. In this paper we propose enhanced firefly algorithm where main improvements are correlated to the hybridization with the exploration mechanism from another swarm intelligence algorithm, introduction of new exploitation mechanism and parameter-based tuning of the exploration-exploitation balance. We tested our approach on the same standard benchmark functions and showed that it not only overcame weaknesses of the original firefly algorithm, but also outperformed other state-of-the-art swarm intelligence algorithms.

This work addresses the target tracking problem that makes use of combined measurements, namely received signal strength (RSS) and angle of arrival (AoA). By linearizing the measurement models and incorporating the prior knowledge obtained from target state transition model, we show that the application of the Kalman filter (KF) to the considered tracking problem is straightforward. Then, an extension of the linearization approach to the case where the target transmit power is not known is introduced and applied to the measurement model to obtain an estimate of the transmit power. By taking advantage of this estimated value, we show that the proposed KF algorithm can easily be generalized to the case of unknown transmit power. Our simulation results confirm the efficacy of the proposed algorithms in comparison with the existing one, as well as the robustness of the proposed approach to not knowing the transmit power. Finally, the supremacy of using the Bayesian approach in comparison with the classical one which disregards the prior knowledge information is also validated through computer simulations.

This paper presents implementation of the fireworks algorithm for portfolio optimization problem with constraints. Fireworks algorithm is a relatively new nature-inspired meta-heuristic which emulates the process of fireworks' explosion. We adapted fireworks algorithm for solving constrained portfolio selection problem that extends classical mean-variance portfolio model by adding additional constraints. Comparative analysis was conducted with three other swarm intelligence algorithms and three variants of genetic algorithm from the literature, using the same problem formulation and the same test data. Results show that the fireworks algorithm has great potential for tackling portfolio optimization problem since it performed better than mentioned algorithms considering all performance indicators.

This paper introduces implementation of hybridized bat algorithm for multi-objective radio frequency identification network planning problem. Multi-objective RFID problem is a well known hard optimization problem that can be solved by using swarm intelligence algorithms. Bat algorithm is a recent mataheuristic, proved to be very successful for tackling such tasks. In our implementation, we hybridized bat algorithm with the artificial bee colony algorithm and adapted it for solving radio frequency identification network planning problem. In the experimental section, we have first shown, by using standard bound-constrained benchmark functions, that our hybridization is justified and that it improves results compared to standard bat algorithm, as well as to other state-of-the-art algorithms. After that, we examined performance of our proposed approach on illustrative RFID network planning problem and compared it with other results from the literature where our proposed algorithm proved to be more successful.