Failure prediction in cloud environment using deep learning

Abstract

In recent years deep learning techniques shown a great results in many fields (such as natural language processing, computer vision). Among their new applications, researchers started to apply these techniques on time series data to predict failures. Predicting failures in cloud environment is a recent research area where researchers try to achieve good results to maintain the service availability of a cloud service. In this project we applied one of deep learning techniques which is neural networks, specifically LSTM on google cluster dataset to predict failures, our model shown good results after extensive evaluation and experimental test.