Automatic Excursion Detection in Manufacturing: Preliminary Results.

摘要: In the past decades, high-volume manufacturing processes have grown increasingly complex. If a failure in these systems is not detected timely manner, it often results tremendous costs. Therefore, demand for methods that automatically detect failures high. this work, we address problem of automatic excursion detection based on parametric tests. Overlooking complexity wafer fabrication processes, propose two structurally simple models: Naive Bayes classifier and boosted decision stumps. We apply models domain semiconductor manufacturing, compare them to offthe-shelf classification techniques, show significant gains precision recall excursions. These encourage our future work should primarily focus increasing models.