An early fault warning freamework for large communication systems

Abstract

With the enormous growth of information in the digital age, especially with the presentation of large systems and cloud computing or inter-cloud computing systems, the work of managing those systems become more and more challenging and time consuming, hence depend deeply on supporting tools. There are several new technique that being develop based on machine learning to help solving that problems. However those machine learning algorithm does still need further investigation since it is not only heavily depends on the characteristic of data, but also the training progress. Hence, it is important to have an overview of evaluation to find which the most suitable model is for these specific dataset. This thesis provide the information on those aspects regarding the evaluation and optimization of those predictive models.