Semi-Supervised Classification of Wafer Map Based on Ladder Network

Wafer map analysis is a key step in semiconductor manufacturing process. The various gross failing area (GFA) patterns on wafer maps are helpful to identify the root causes of failures. In this work, a semi-supervised classification framework is proposed for wafer map analysis. We use inline defect wafer map as the example, especially for GFA pattern classification. After data preprocessing and selection, Ladder network is adopted here to classify wafer maps compared with a standard convolutional neural network (CNN) model on two real-world datasets. The results illustrate that Ladder network is consistently and substantially better than CNN model across various training data percentages by effective utilization of the unlabeled data.