Application of Neural Networks to Optimize Distributed Computing in Cloud and Edge Environments

This study addresses the pressing challenges of resource optimization in distributed cloud and edge computing environments. We propose a deep learning-based approach utilizing Long Short-Term Memory (LSTM) neural networks for accurate server load forecasting and dynamic resource management. By analyzing historical workload data, the model predicts future demand peaks and supports proactive scaling decisions. A synthetic time-series dataset was developed to simulate realistic workload fluctuations, including periodic patterns, noise, and anomalies. Experimental results demonstrate that the LSTM model achieves high accuracy in capturing workload dynamics and significantly outperforms traditional heuristic approaches in terms of load balancing efficiency and energy consumption. Despite occasional limitations in forecasting sudden load spikes, the proposed neural network architecture proves to be a robust solution for intelligent resource allocation in real-time distributed systems. The findings highlight the potential of deep learning to enhance adaptability, responsiveness, and sustainability in modern computing infrastructures.