Deep Learning-Based Resource Allocation and Task Scheduling in Cloud Computing: A Review

Abstract

Cloud computing has emerged as a dominant paradigm for delivering scalable, on-demand computing resources to users across diverse domains. However, efficient resource allocation and task scheduling remain critical challenges due to the dynamic, heterogeneous, and distributed nature of cloud environments. Traditional scheduling approaches often fail to handle the increasing complexity, leading to issues such as resource underutilization, high latency, and increased operational cost. Recent advancements in deep learning and optimization techniques have provided promising solutions to these challenges. In particular, hybrid approaches combining neural networks with metaheuristic optimization algorithms have demonstrated improved performance in handling multi-objective optimization problems. This review focuses on recent developments in deep learning-based resource allocation and task scheduling, with special emphasis on hierarchical auto-associative polynomial convolutional neural networks (HAPCNN). These models enable efficient feature extraction, adaptive learning, and dynamic resource management in cloud systems. Furthermore, optimization techniques such as reinforcement learning, swarm intelligence, and evolutionary algorithms are explored for enhancing scheduling efficiency. The paper systematically analyses recent studies, identifies research gaps, and highlights future directions. The findings indicate that integrating deep learning with optimization frameworks significantly improves Quality of Service (QoS), reduces execution time, and enhances resource utilization in cloud computing environments.