Optimizing Failure Prediction in Cloud Computing Using LSTM and MLP Integration

Accurate failure prediction in cloud computing is vital for maintaining system reliability and minimizing downtime. This paper presents an optimized approach for cloud failure prediction by integrating Long Short-Term Memory (LSTM) and Multi-Layer Perceptron (MLP) models, leveraging the Google Cloud Trace dataset. To enhance the predictive accuracy and efficiency of the models, Grey Wolf Optimization (GWO) algorithm is employed for feature selection, ensuring that the most relevant features are utilized in the prediction process. The integration of LSTM's temporal sequence learning capabilities with MLP's deep learning strengths enables a robust prediction framework that adapts to complex patterns in cloud environments. Experimental results demonstrate that the proposed method significantly improves failure prediction accuracy compared to traditional approaches, while the GWO-based feature selection contributes to reduced computational overhead and enhanced model performance. This study highlights the effectiveness of combining advanced machine learning techniques with optimization algorithms to address the challenges of failure prediction in cloud computing.