Machine Learning Algorithms for Fault- Prediction

Abstract

With the advancement in VLSI technology, the number of transistors on a device increases along with the reduction in the size of transistor. The likelihood of a manufacturing failure rises as feature sizes continue to contract. The overall testing cost and testing efforts are increases exponentially with each new technology node. Therefore, it is necessary to explore the techniques which guarantee the circuit functioning with less efforts and cost. As with each new technology node, not only the possible number of faults in circuit increases but also new types of faults are being introduced. In this scenario, this paper aims to explore the various existing Machine Learning (ML) methods for the prediction of number of faults in circuit. This paper also aims to categorize the fault prediction and prediction of test vector set. The paper includes the comparison analysis of different ML algorithms in fault prediction. With the use of ML algorithm, the Automatic Test Pattern generator (ATPG) shortens the time needed to generate test set required for manufacturing testing.