Recipe Generation from Small Samples: Incorporating an Improved Weighted Kernel Regression with Correlation Factor

The cost of the experimental setup during the 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 consist of only a few samples. As a result, existing machine learning algorithms cannot be applied in this setting. To solve this problem, predictive modeling algorithm called Weighted Kernel Regression with correlation factor (WKRCF), which is based on Nadaraya-Watson kernel regression (NWKR), is proposed. The correlation factor reflected the important features by changing the bandwidth of the kernel as a function of the output. Even though only four samples are used during the training stage, the WKRCF provides an accurate prediction as compared with other techniques including the NWKR and the artificial neural networks with back-propagation algorithm (ANNBP). Thus, the proposed approach is beneficial for recipe generation in an assembly process development.