IC Test Quality Enhancement by Introducing Machine Learning

“Big data” is a popular term everywhere. However, it becomes another topic how to take advantage of data and induction/deduction effective conclusions to support quality enhancement. In industry, big data accompanies with big effort due to manufacture complex architectures. People still frequently fine tune rules by manual, either most of academic researches not able to fit complicated semiconductors process due to exceptions. Data is correct but lack of industrial knowledge base. In this paper, we pointed out IC test quality bottlenecks in data confuse which influences machine learning analysis and embedded adaptive model from die, lot and product levels on simple workable mechanism by vertical and horizontal two machine learning dimensions to make system “like live” in automatically ways to enhance test quality.