Recipe generation from small samples by weighted kernel regression
The cost of experimental setup during an assembly process development of a chipset, particularly the under-fill process, can often result in insufficient data samples. In INTEL Malaysia, for example, the historical chipset data from an under fill process consists of only a few samples. As a result,...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Book Section |
| Published: |
IEEE
2011
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/29575/ http://dx.doi.org/10.1109/ICMSAO.2011.5775473 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The cost of experimental setup during an assembly process development of a chipset, particularly the under-fill process, can often result in insufficient data samples. In INTEL Malaysia, for example, the historical chipset data from an under fill process consists of only a few samples. As a result, existing machine learning algorithms for predictive modeling cannot be applied to this setting. Despite this challenge, the use of data driven decisions remains critical for further optimization of this engineering process. In this study, a weighted kernel regression (WKR) is introduced to improve the predictive modeling in the setting with limited data samples. In the proposed framework, the original Nadaraya-Watson kernel regression (NWKR) algorithm is modified. Even though only four samples are used during the training stage of our experiment, the proposed approach is able to provide an accurate prediction within the engineer’s requirements as compared with other existing predictive modelings including NWKR and artificial neural networks with back-propagation algorithm (ANNBP). Thus, the proposed approach is beneficial for recipe generation in an assembly process development. |
|---|