Overview / Usage

The reliability of a software is the amount of time for which the software continuously operates. This project involves the temporal modeling of the Time Between Failure (TBF) series for different software applications, using advanced machine learning algorithms like MOGA-NN (Multiobjective Genetic Algorithm) and ELM (Extreme Learning Machines). For generalizing our results, we performed our experiments on three different types of software systems: Real Time Systems, Military Systems and Word Processing Systems.

Methodology / Approach

The Multiobjective Genetic Algorithm uses NSGA-II for optimising the cost given by the neural network, which is trained to provide the prediction intervals for the time-between-failure of the test data.

The Extreme Learning Machines are combined with kNN approach to get the prediction interval.