Cloud computing primarily focuses on providing secure and reliable access to geographically distributed resources. Cloud computing is a parallel and distributed system that simplifies the virtualization of distributed computing. Job scheduling is one of the most important services in cloud computing. Job scheduling is used to schedule users' jobs to allocate suitable resources in a cloud environment. Recently, several efficient job scheduling algorithms have been proposed for cloud computing. The purpose of designing these algorithms is to achieve time and cost optimization for job execution. On the other hand, it is necessary to provide a model to assess the reliability of the different job scheduling algorithms. Unfortunately, no accurate and complete model has been proposed to evaluate the reliability of these algorithms. Scheduling failures are inevitable in cloud computing due to the harsh deployment environment, resource migration, non-compliance time or cost, etc. Therefore, providing a reliability model is necessary for successful scheduling algorithms. In this paper, a reliability model based on a discrete-time Markov chain is incorporated for the famous scheduling algorithms MCT, MET, BCO, HCOC, and PPO. We propose an approach to evaluating the reliability of scheduling algorithms based on time and cost constraints. The proposed approach is modeled by a Discrete Time Markov Chain, and the results show the reliability of scheduling algorithms in different states.

Wireless Sensor Networks (WSNs) face significant challenges due to limited energy resources and the need for optimal resource allocation. One of the key issues in these networks is efficient energy management and intelligent allocation of bandwidth and transmission power to enhance network lifetime while maintaining Quality of Service (QoS). Traditional resource allocation methods often fail to achieve optimal performance because they cannot adapt to network conditions or accurately assess node inefficiencies. Therefore, developing intelligent, adaptive approaches to resource management in WSNs is crucial. The proposed method consists of three phases: leveraging Data Envelopment Analysis (DEA), deep learning-based evolutionary optimization, and adaptive network management for optimal resource allocation in WSNs. In the first phase, the Slack-Based Measure (SBM) DEA model is employed to assess the efficiency of sensor nodes. This model evaluates inefficiencies by considering Slack values and clusters sensor nodes based on energy consumption, delay, bandwidth, and data transmission rate. The output of this phase identifies inefficient nodes, which are then forwarded to the second phase for further optimization. In the second phase, a Deep Neural Network (DNN) is employed to allocate resources to inefficient nodes. The DNN is trained via standard backpropagation on the output data from the first phase to predict the optimal transmission power and bandwidth allocation. The Whale Optimization Algorithm (WOA) is used exclusively to optimize the DNN hyperparameters, with the objective of minimizing the overall energy consumption. In the third phase, adaptive resource management updates resource allocation in response to changing network conditions, such as congestion, delay, and QoS fluctuations. This phase prevents excessive congestion, reduces latency, and ensures fairness in resource distribution among nodes. Simulation results demonstrate that the proposed method outperforms existing approaches, achieving 2

Energy-efficient task scheduling is a critical challenge in heterogeneous cloud data centers, where makespan, energy consumption, runtime, and resource feasibility must be optimized simultaneously. This paper proposes a constraint-aware hybrid metaheuristic framework for multi-objective cloud task scheduling by integrating Genetic Algorithm (GA), Grey Wolf Optimizer (GWO), threshold-based local refinement, and Colored Petri Net (CPN)-based event-driven simulation. The CPN module formally models task dispatching, parallel execution, CPU, RAM, and bandwidth contention, resource allocation, task completion, and waiting behavior, enabling feasibility-aware evaluation under concurrent execution. GA and GWO first explore the scheduling search space, and the best solutions are then refined through a threshold-guided local improvement mechanism without relying on penalty functions. Experimental results show that the proposed Hybrid method outperforms GA and GWO, achieving the lowest mean fitness of 674.20, makespan of 821.96, energy consumption of 452.56, and runtime of 10.20. It also records the smallest fitness standard deviation of 25.22, confirming higher stability. Statistical tests confirm significant improvements, with p-values of 1.024455e-07 against GA, 1.862645e-09 against GWO, and 2.066489e-10 for the Friedman test. Ablation, convergence, sensitivity, Pareto, ranking, energy, and scalability analyses further verify its robustness.

Deep medical clustering aims to discover latent disease subgroups from unlabeled medical data. However, most existing approaches produce deterministic cluster assignments and provide limited mechanisms for calibrated uncertainty communication or clinical interpretation. This paper proposes CUPID-Med, a conformal uncertainty-aware prototype discovery framework for clinically interpretable deep medical clustering. CUPID-Med integrates medical foundation encoders, distributional latent representation learning, prototype-based deep clustering, conformal uncertainty calibration, and clinically interpretable outputs. Rather than assigning every sample to a single hard cluster, CUPID-Med generates calibrated set-valued cluster predictions and flags ambiguous cases for expert review. The framework was evaluated on a 5,000-image frontal CheXpert subset using a no-label training protocol, where disease labels were excluded from training and used only for post-hoc clinical interpretation. Preliminary clustering baselines applied to DenseNet121 embeddings produced weak separation, with the best baseline Silhouette score of 0.0510. In contrast, the primary CUPID-Med run achieved a Silhouette score of 0.5175, a Davies--Bouldin index of 2.7809, and a Calinski--Harabasz score of 29,733.16, while also producing prototype assignments, confidence estimates, and latent uncertainty scores. These results suggest that distributional prototype learning can produce more structured and interpretable medical image subgroups than deterministic clustering of fixed deep embeddings.

Clustering high-dimensional biomedical data without prior knowledge of the number of clusters remains a major challenge in medical image and signal analysis. We present SONSC (Separation-Optimized Number of Smart Clusters), an adaptive and interpretable clustering framework driven by the Improved Separation Index (ISI)—a novel internal validity metric that jointly evaluates intra-cluster compactness and inter-cluster separability. SONSC iteratively maximizes ISI across candidate cluster configurations to automatically infer the optimal number of clusters, without supervision or parameter tuning. Extensive experiments on benchmark datasets (MNIST, CIFAR-10) and real-world clinical modalities (chest X-ray, ECG, RNA-seq) demonstrate that SONSC consistently outperforms classical methods such as K-Means, DBSCAN, and spectral clustering in ISI, Silhouette score, and normalized mutual information (NMI). Beyond numerical performance, SONSC identifies clinically coherent structures aligned with expert-labeled categories, supporting its integration into diagnostic and patient stratification pipelines. By unifying algorithmic robustness with medical interpretability, SONSC provides a scalable and trustworthy solution for unsupervised biomedical data analysis.

Rapid vehicle movement and changes in overall vehicle density and distribution can make routing inefficient, cause delays, and create cluster instability while using VANETs (Vehicular Ad Hoc Networks). This paper intelligent frameworks focus on enhancing cluster stability and divides it into two stages: (1) the sophisticated weighted K-Medoid algorithm for signal strength and relative speed, node density, and movement direction-based initial clustering (2) dynamic weight adjustment self-organizing map (SOM) cluster head selection reinforcement learning (RL) optimization. NS-3 simulations (node density 50–150, speed 10–30 m/s) demonstrate the frameworks Adaptive RL border patrol outperformed SDPC, RL-neighbor selection, K-means, fuzzy logic with 9% SDPC, 15% K-means more PDR, improved end-to-end delay, cluster linger time, control traffic and greater RC. Adding RL improved adaptability and stabilization in high mobility fluctuating mover scenarios.

This work introduces the “MedTransCluster” framework, a novel approach to medical image clustering in chest radiography through the application of transfer learning, leveraging the capabilities of pre-trained deep learning models. Our evaluation encompassed a variety of neural networks, considering their adaptability to the nuances of medical imaging data. The study incorporated four renowned clustering algorithms and an expanded set of evaluation metrics, offering a comprehensive comparison and a refined analysis of these models’ ability to cluster complex diagnostic features. Notably, EfficientNetB0 coupled with DBSCAN clustering algorithm achieved a silhouette score of 0.924131, and ResNet152 with KMeans displayed a Calinski Harabasz score of 9655.213964, indicating their superior proficiency in capturing the intricacies of medical features. These results emphasize the critical importance of model refinement within the healthcare imaging sphere and underscore the potential of methodologies like MedTransCluster in enhancing diagnostic accuracy and patient outcomes.

Computers, multiprocessor platforms on a chip (MPSoC) and embedded systems have many applications in the military, security, medical, and space industries. Improving reliability and trust in these systems is very important. By improving reliability in multiprocessor and embedded systems, their efficiency and performance will increase. Multi-processor platforms in embedded systems (EMMPSoC) can perform sensitive and critical tasks by communicating with each other; But in the execution process, they may encounter errors and problems in the processing. The existence of errors and problems in the execution process, temperature and heat generation, high energy consumption, unfavorable execution time, etc. in these systems leads to a decrease in reliability and the occurrence of irreparable events. One of the most important challenges in MPSoC is improving reliability. The reliability of these systems can be increased by estimating the Mean Time to Failure (MTTF). In this research, a Self-directed Model based-on Clipping Double Q-Learning (CDQL) and extended Markov model (EMM) and meta-heuristic MUO algorithm (Self -QMM) to estimate the average failure time in EMMPSoC systems. has been raised The CDQL has been used to predict the critical time in MPSoC systems and the developed Markov model has been used to select the appropriate processor. By simulating the proposed Self-QMM model, it has been improved that the energy consumption in the Self-QMM model is 50, 27 and 14 kJ, respectively, compared with the latest proposed methods such as FIS, FNN-NSGA-II and MUO method, the temperature production rate and temperatures have improved by 45.4, 22.9 and 9.9 degrees Celsius and the execution time by 0.16 minutes (9.6 s), 0.09 minutes (5.4 s) and 0.08 minutes (4.8 s).

The present paper proposes a novel algorithm with the highest accuracy and efficiency for resource allocation in the IoT, considering the fog computing technique. To meet the need for resource allocation, including central processing unit, bandwidth, and main memory, the Genetic Algorithm, Particle Swarm Optimization, Adaptive Weighted Particle Swarm Optimization, Crystal Structure Algorithm, and Simulated Annealing were examined. The obtained results emphasize the better performance of Simulated Annealing compared to the rest. The findings of this paper can benefit optimizing the tasks migration in internet of things. The major innovation of this research lies in the heart of the comparison made between the selected algorithms to see which one conducted the migration tasks with the highest accuracy and remarkable convergence rate.

There have been many connections between medical science and artificial intelligence in recent years. Many problems arise with the integrity of communication. Cardiac arrhythmia, carried out using artificial intelligence methods, is one of the most dangerous diseases in the field of prevention. Topics introduced in artificial intelligence are the automatic selection of balancing and classification algorithms. In this study, metrics for machine learning algorithm selection are presented. The first problem is the problem of choosing the best balancing algorithm to balance the data sets, introduced as triangle rate (TR). The second issue to be studied is selecting the best automatic classification algorithm. The third action was to use a scoring algorithm to predict sinus and non-sinus arrhythmias. The heptagonal reinforcement learning (HRL) achieved results competitive with standard algorithms by combining three types of algorithms. The data used in this study was a 12-lead electrocardiogram (ECG) database of arrhythmias. The number of patients examined in this dataset is 10,646. The HRL algorithm has improved the previous algorithms by 5

Due to the completely random and dynamic nature of the cloud environment, as well as the high volume of jobs, one of the significant challenges in this environment is proper online job scheduling. Most of the algorithms are presented based on heuristic and meta-heuristic approaches, which result in their inability to adapt to the dynamic nature of resources and cloud conditions. In this paper, we present a distributed online algorithm with the use of two different learning automata for each scheduler to schedule the jobs optimally. In this algorithm, the placed workload on every virtual machine is proportional to its computational capacity and changes with time based on the cloud and submitted job conditions. In proposed algorithm, two separate phases and two different LA are used to schedule jobs and allocate each job to the appropriate VM, so that a two phase adaptive algorithm based on LA is presented called TPALA. To demonstrate the effectiveness of our method, several scenarios have been simulated by CloudSim, in which several main metrics such as makespan, success rate, average waiting time, and degree of imbalance will be checked plus their comparison with other existing algorithms. The results show that TPALA performs at least 4.5% better than the closest measured algorithm.

Amidst the surge of Big Data, deep clustering emerges as a pivotal technique in machine learning, necessitating robust and interpretable evaluation metrics that align with its complexities. Traditional metrics, largely dependent on Euclidean distance, falter in capturing the essence of deep clustering in high-dimensional data. This paper proposes the Cosine Clustering Index (CCI), an innovative metric that leverages cosine distance to accurately assess clustering outcomes in deep learning contexts. Distinguished by its ability to evaluate intra-cluster cohesion and inter-cluster separation, CCI addresses the intricacies of deep clustering more effectively than existing metrics. Validation on benchmark datasets like IRIS, MNIST, and CIFAR10 showcases CCI’s enhanced interpretability, scalability, and its superior adaptability to the depth and dimensionality of data. The introduction of CCI marks a significant stride towards advancing clustering evaluation, promising a more profound understanding and analysis of deep clustering mechanisms.

This study investigates the application of heuristic and meta-heuristic algorithms to address resource allocation challenges in Internet of Things (IoT) applications within fog computing environments. The primary advantage of these algorithms lies in their ability to optimize functions without the need for stringent restrictions, allowing adaptability to various linear, nonlinear, continuous, or discrete problems. Through the implementation and comparison of the Cheetah algorithm, Gray Wolf algorithm, Particle Swarm-Gravitational Search algorithm, and Gray Wolf-Cuckoo Search algorithm using MATLAB software in a simulation environment, the study aims to minimize criterion function and total time and energy consumption for IoT applications. Preliminary results indicate that the statistical average performance of the Cheetah algorithm surpasses that of the Gray Wolf algorithm, the combined Particle Swarm-Gravitational Search algorithm, and the Gray Wolf-Cuckoo Search algorithm. This suggests the efficacy of the Cheetah algorithm in IoT resource allocation optimization within fog computing environments. The study provides insights into the comparative performance of these algorithms, laying the foundation for further exploration into enhancing resource allocation strategies in the dynamic and resource-constrained IoT and fog computing landscapes.

Prioritization of activities is a multicriteria problem that includes both quantitative and qualitative factors. Moreover, due to the nature of organizations and activities, the impact of activities on each other is fuzzy. Prioritizing activities in real and fuzzy situations will help an organization’s decision makers to make the right decisions. In this paper, a new fuzzy hybrid methodology is proposed to describe and prioritize the activities of the organization in fuzzy conditions. First, the activities of the organization are described in a new fuzzy format based on activity model (OV-5) of the C4ISR framework. Furthermore, fuzzy DEMATEL is used to calculate the interdependencies between activities, and then fuzzy ANP is used to calculate the weight of each activity. Finally, by combining the results of the two methods, the priority of activities was calculated. Additionally, an empirical study is presented to illustrate the application of the proposed method.

Wireless sensor networks (WSNs) are very important to realize the Internet of Things by connecting physical objects to the Internet. RPL (Routing Protocol for Low Power and Lossy Networks) has been provided by the IETF (Internet Engineering Task Force) for routing on wireless sensor networks. RPL uses the Trickle algorithm to schedule the transmission of control messages. The environment of WSNs is often highly variable and the environmental conditions for nodes are different. If network nodes are equipped with a learning automaton, network convergence time can be reduced. The use of learning automata allows each network node to adjust its parameters to environmental conditions by receiving environmental feedback to perform better. The main aim of this article is to reduce the network convergence time. We equipped the Trickle algorithm with a learning automaton, which determines how many times the algorithm is repeated with the minimum interval to resolve the inconsistency, based on environmental conditions. Since repeating the Trickle algorithm with the minimum interval increases the local repair speed and reduces the network convergence time. We simulated a large number of networks of different sizes and densities. According to the simulation results, we observed that in the proposed method, the network convergence time was reduced compared with the other methods. Also, due to fewer changes in network topology, energy consumption for network reconstruction was reduced.

Nowadays, artificial intelligence is widely used in various fields and industries. Cybercrime is a concern of these days, and artificial intelligence is used to detect this type of crime. Crime detection systems generally detect the crime by training from the related data over a period of time, but sometimes some samples in a dataset may have no label. Therefore, in this paper, a method based on semi-supervised neural network is presented regarding crime types detection. As the neural network is a supervised classification system, therefore, this paper presents a pseudo-label method for neural network optimization and develops it to semi-supervised classification. In the proposed method, firstly the dataset is divided into two sections, labelled and unlabelled, and then the trained section is used to estimate the labelling of the unlabelled samples based on pseudo-labels. The results indicate that the proposed method improves the accuracy, Precision and Recall up to 99.83%, 99.83% and 99.83%, respectively.

The main purpose of this research is to model the process and extract patterns of computer crime using data mining and employing MATLAB software for data modeling. The results have been simulated and presented graphically. The simulation results show that this system can be considered as one of the most effective and lowest cost ways to identify the cyber-criminal behavior, therefore, computer crime experts can run effectively this model on their systems.