Recurrent neural network for software failure prediction

Abstract

Software failure occurs when the software runs in an operational profile. Controlling failures in software require that one can predict problems early enough to take preventive action. The prediction of software failures is done by using the historical failures collected previously when they occur. To predict software failures, several models are proposed by researchers. In this paper, we present a recurrent neural network (RNN) to predict software failure using historical failure data. The proposed RNN is trained and tested using collected data from the literature; the obtained results are compared with other models and show that our proposed model gives very attractive prediction rates.