Abstract:

The scale and complexity of cloud computing infrastructures make efficient resource allocation management a critical problem in system stability, cost effectiveness, and optimal performance. This research presents a new approach that aims to optimize scheduling and resource allocation in cloud environments through neural networks and reinforcement learning. The proposed framework uses reinforcement learning alongside neural networks to predict workload requirements, allowing dynamic shifts in resource allocation based on system status. Results show superior performance of the hybrid model over individual ML-based approaches and traditional heuristics in resource utilization, reduction of latency, and reduction in operational costs. The results illustrate the integration of deep learning and decision-making algorithms to adapt to the requirements of cloud computing systems.