Explore the vast range of titles available.

Marine fishery resource prediction is crucial for sustainable fishery management and ecosystem conservation, yet traditional statistical methods face limitations in capturing the complex non-linear relationships and multi-scale temporal dependencies inherent in marine environmental systems. This study proposes a novel CNN-XGBoost fusion model that integrates convolutional neural networks’ temporal pattern recognition capabilities with extreme gradient boosting’s ensemble learning strengths for enhanced marine fishery resource forecasting. The fusion architecture employs a hierarchical two-stage framework where CNN components extract high-level temporal features from multi-source marine environmental data, while XGBoost modules process both extracted features and engineered variables to generate final predictions. Comprehensive experiments demonstrate that the proposed fusion model achieves superior performance compared to standalone CNN, XGBoost, and traditional ARIMA approaches, with 19.1% improvement in RMSE and statistically significant enhancements across all evaluation metrics. The optimal fusion weight analysis reveals that CNN-extracted features and XGBoost-processed features are weighted at 40 and 60% respectively in the final prediction fusion, achieving RMSE of 2.847, MAE of 2.184, and R 2 of 0.846. These percentages represent fusion weight allocation rather than prediction accuracy values. Time series analysis confirms robust performance across seasonal variations and exceptional capability in predicting extreme abundance events critical for adaptive fishery management. The results provide valuable insights for sustainable marine resource management and offer practical tools for fishery policymakers and resource managers.

Load balancing (LB) is a critical aspect of Cloud Computing (CC), enabling efficient access to virtualized resources over the internet. It ensures optimal resource utilization and smooth system operation by distributing workloads across multiple servers, preventing any server from being overburdened or underutilized. This process enhances system reliability, resource efficiency, and overall performance. As cloud computing expands, effective resource management becomes increasingly important, particularly in distributed environments. This study proposes a novel approach to resource prediction for cloud network load balancing, incorporating federated learning within a blockchain framework for secure and distributed management. The model leverages Dilated and Attention-based 1-Dimensional Convolutional Neural Networks with bidirectional long short-term memory (DA-DBL) to predict resource needs based on factors such as processing time, reaction time, and resource availability. The integration of the Random Opposition Coati Optimization Algorithm (RO-COA) enables flexible and efficient load distribution in response to real-time network changes. The proposed method is evaluated on various metrics, including active servers, makespan, Quality of Service (QoS), resource utilization, and power consumption, outperforming existing approaches. The results demonstrate that the combination of federated learning and the RO-COA-based load balancing method offers a robust solution for enhancing cloud resource management.

High-end data centers are required to process the user requests and provide them with a better quality of service. The prominent issues in building a sustainable data center are reduced carbon footprint, dynamic capacity planning to reduce resource provisioning time and cost, minimized virtual machine migration to prevent higher downtime and enhanced return on investment and resource utilization. Realizing true elasticity will be a solution for these issues. Better elasticity can result if the data center is aware of the workload before its entry. Hence, the data center has to have a predictive model to forecast the resource requirements before the arrival of the workload. We propose a novel methodology called clustering-cum-Markov to predict the workload resource requirements proactively. It runs in the data center’s controller and collects the statistics of the incoming workload. It characterizes the workload and predicts the necessary resources two-time slots ahead. We evaluate the modle in our data center and also with the benchmark Google Workload dataset. The results are compared with the state-of-the-art solutions based on various metrics, including the environment metrics. The proposed model achieves a 99.01% precision and exhibits optimal values with respect to the environmental metrics.

With the continuous growth of cloud computing and virtualization technology, network function virtualization (NFV) techniques have been significantly enhanced. NFV has many advantages such as simplified services, providing more flexible services, and reducing network capital and operational costs. However, it also poses new challenges that need to be addressed. A challenging problem with NFV is resource management, since the resources required by each virtualized network function (VNF) change with dynamic traffic variations, requiring automatic scaling of VNF resources. Due to the resource consumption importance, it is essential to propose an efficient resource auto-scaling method in the NFV networks. Inadequate or excessive utilization of VNF resources can result in diminished performance of the entire service chain, thereby affecting network performance. Therefore, predicting VNF resource requirements is crucial for meeting traffic demands. VNF behavior in networks is complex and nonlinear, making it challenging to model. By incorporating machine learning methods into resource prediction models, network service performance can be improved by addressing this complexity. As a result, this paper introduces a new auto-scaling architecture and algorithm to tackle the predictive VNF problem. Within the proposed architecture, there is a predictive VNF auto-scaling engine that comprises two modules: a predictive task scheduler and a predictive VNF auto-scaler. Furthermore, a prediction engine with a VNF resource predictor module has been designed. In addition, the proposed algorithm called GPAS is presented in three phases, VNF resource prediction using genetic programming (GP) technique, task scheduling and decision-making, and auto-scaling execution. The GPAS method is simulated in the KSN framework, a network environment based on NFV/SDN. In the evaluation results, the GPAS method shows better performance in SLA violation rate, resource usage, and response time when compared to both ElasticSFC and Basic methods (without any auto-scaling algorithm). Furthermore, in terms of the resource prediction model, the GP method employing the ANN regressor has demonstrated superior outcomes compared to alternative methods that were tested. Considering the results, it can be concluded that resource prediction in task scheduling and decision-making in advance is an effective and efficient approach to run VNF auto-scaling.

This study aimed to predict the lithium resource potential in the Jiulong region of western Sichuan using a spectral residual attention convolutional neural network (SRACN) model, which integrates hyperspectral imagery from the GF-5B satellite with spectral measurement data from field rock core samples. By incorporating residual connections and a spectral attention mechanism, the SRACN model efficiently extracts critical spectral features, thereby enhancing mineral identification accuracy and predictive performance. The experimental results demonstrated that: (1) The SRACN model achieved a classification accuracy of 96.46% and an F1 score of 0.9645 for muscovite classification and mineral mapping, indicating superior performance; (2) utilizing hierarchical density-based spatial clustering of applications with noise (HDBSCAN), lithium and rare metal mineralization zones in the Jiulong region were delineated, with results closely aligned with field validation, revealing significant exploration potential in the northern Daqianggou mining area and the Baitaizi region. This study presents a novel scientific and technical approach to regional geological prospecting and demonstrates the effectiveness of integrating SRACN with density clustering analysis for evaluating regional mineral resource potential.

Traditional cloud architectures struggle to effectively allocate resources to container-based workloads due to fluctuating usage patterns and potential interference among multi-tenants. Conventional scheduling methods, which primarily rely on user-specified resource requests, often lead to over-provisioning and suboptimal resource utilization. Although efforts have been made to predict container resource usage and allocate resources more tightly than the full requests, such approaches typically fall short during sudden demand spikes, thus failing to meet Service Level Objectives (SLOs). In this article, we introduce a novel cloud resource prediction engine specifically designed to differentiate between online and batch jobs. Our engine prioritizes ensuring SLOs for online jobs where immediate responsiveness is crucial. Specifically, our approach employs a combination of genetic algorithms (GA) and Bayesian neural networks (BNN) to enhance the prediction accuracy of CPU and memory resources. Trained on real-world trace data, our model significantly outperforms traditional forecasting methods like ARIMA and exponential smoothing, especially in reducing the risk of under-prediction for online jobs. This not only ensures more efficient resource utilization but also improves adherence to SLOs without compromising performance.

The deployment of industrial robots in time‐critical applications demands ultra‐low latency and high reliability in communication systems. This study presents a novel delay optimisation framework for industrial robot control systems using 6G network slicing technologies. A Gale–Shapley (GS)‐based elastic switching model is proposed to dynamically match robot controllers to optimised network slices and base stations under latency‐sensitive conditions. To enhance resource adaptability, a long short‐term memory (LSTM)‐based encoder‐decoder structure is developed for predictive resource allocation across slices. The proposed integrated matching mechanism achieves a success rate of 91.16% for slice access and a base station access rate of 90.83%, outperforming conventional integrated and two‐stage schemes. The LSTM‐based resource allocation achieves a mean absolute error of 0.04 and a violation rate below 10%, with over 92% utilisation of both node and link resources. Experimental simulations demonstrate a consistent end‐to‐end latency below 7 ms and a throughput of 18.4 Mbit/s, validating the proposed models' effectiveness in ensuring robust, real‐time communication for industrial robot operations. This research contributes a scalable solution for dynamic 6G network resource management, providing a foundation for advanced industrial automation and intelligent manufacturing. A novel elastic switching model based on the Gale–Shapley (GS) algorithm and a resource allocation model grounded in an LSTM encoder‐decoder structure, tailored for 6G network slicing scenarios. Through extensive simulations, our model demonstrates a 91.16% slice access success rate, average latency below 5 ms, and resource utilisation exceeding 92%, outperforming conventional integrated matching and static allocation methods.

Accurate prediction of cloud resource instances is becoming increasingly important for public cloud users and cloud service providers, because it touches on the reasonable reservation of cloud resources with minimize costs. However, current methods do not predict the instance types of cloud resources based on the application workloads from users, and less consider the characteristics of workload data changes in the real-time prediction. To solve these problems, this paper proposes an application workload-dependent cloud resource instance prediction model to predict appropriate cloud instance resource usage in a timely manner. Firstly, we adopt a trend degree (TD) to classify all requested workloads into three types of wave trend patterns. Next, a Hidden Markov model based cloud resource prediction method (HMM-CPM) tracing the requested workload trends is presented. Finally, the reasonable cloud instance types following the patterns of the requested workloads can be predicted. The simulation results show that the proposed method can predict cloud resource instance types in the scenario with certain workload fluctuation, and the prediction accuracy is higher than the existing related approaches.

Nowadays, Cloud services are gaining importance among users due to their cost-effectiveness and highly scalable resources. To meet the user’s demands, several data centres are built across the globe, which has severe environmental as well as economical concerns. Energy consumption is one of the most significant issue faced by cloud service providers. Prediction of accurate resource usage of the physical machine helps in effective utilization of resources in a data centre, resulting in minimizing an active number of physical machines, which helps to minimize the energy consumption of a data centre. Although several models till date focus on virtual machine consolidation with a notion of reducing energy consumption, the reduction of operational physical machines has not gathered enough attention. In this paper, we propose a prediction model to predict resource utilization of physical machines, which enables to effectively utilize the entire data centre’s resources to reduce energy consumption. First, the raw time series workload is processed to enhance the value of its features for better training and prediction of mean resource utilization in the cloud data centre using the proposed Sum Average (SA) algorithm. Afterward, Gaussian Mixture Model (GMM) is employed to cluster heterogeneous machines of data centre based on its resource usage which helps to analyze the prediction for each kind of configured machine available in a data centre. In addition, the Long Short Term Memory model (LSTM) is employed to predict the mean resource usage of physical machines for every clustered machine. Furthermore, the effectiveness of our proposed model is evaluated using the Google cluster trace usage dataset. Lastly, the proposed model is compared with Linear Regression, Moving Average, and Auto Regression Integrated Moving Average model. Root Mean Square Error (RMSE) analysis states that our proposed model outperforms the other compared techniques