Answers are grounded in this paper and cite their sources.
Cloud infrastructures require intelligent workload distribution mechanisms to maintain optimal throughput, control power consumption, and ensure efficient resource utilization. Traditional load balancing approaches fail to address several persistent issues - including workload unpredictability, delayed response cycles, server overutilization, and suboptimal task scheduling. This study proposes an Adaptive Bayesian Belief Network-Ridge Regression framework, termed MRA-SAO-ABBNRR, which employs a Modified Random Attribute Snow Ablation Optimizer. The proposed model integrates Bayesian Belief Networks with Ridge Regression to forecast future workload patterns from historical cloud telemetry. Overfitting is mitigated through correlation analysis across input attributes. Additionally, a modified random attribute snow ablation optimization scheme is used to fine-tune workload prediction and to support a multi-objective task allocation strategy spanning resource utilization, response time, makespan, scalability, and energy efficiency. All experiments were conducted in MATLAB 2020a, and benchmark comparisons were drawn against four existing models, namely Convolutional Neural Network-Long Short-Term Memory, Variable Model Matching Intelligent Self-Distillation, Ensemble Prediction and Risk Uncertainty Assessment, and Bayesian Belief Networks-Ridge Regression. The proposed framework delivered superior prediction accuracy alongside more effective load balancing. It recorded the lowest SMAPE of 0.2 when paired with the sigmoid activation function, outperforming every baseline. The model also demonstrated reduced utilization of compute resources, lower operational expenditure, and decreased energy demand.
No takes yet. Share an insight, caveat, or question.
Punna et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: