Semiconductor fabrication facility design using a hybrid search methodology
Computers and Industrial Engineering
Feature Selection for Knowledge Discovery and Data Mining
Feature Selection for Knowledge Discovery and Data Mining
Data mining for yield enhancement in semiconductor manufacturing and an empirical study
Expert Systems with Applications: An International Journal
Recognition of semiconductor defect patterns using spatial filtering and spectral clustering
Expert Systems with Applications: An International Journal
Defect spatial pattern recognition using a hybrid SOM-SVM approach in semiconductor manufacturing
Expert Systems with Applications: An International Journal
| Hi-index | 0.00 |
Because many factors are complexly involved in the production of semiconductors, semiconductor manufacturers can hardly manage yield precisely. We present a hybrid machine learning system, i.e., a clustered feature-weighting case-based reasoning, to detect high-yield or low-yield lots in semiconductor manufacturing. The system uses self-organizing map neural networks to identify similar patterns in the process parameters. The trained back-propagation neural networks determine feature weights of case-based reasoning. Based on the clustered feature-weighting case-based reasoning, the hybrid system predicts the yield level of a new manufacturing lot. To validate the effectiveness of our approach, we apply the hybrid system to real data of a semiconductor company.