Abstract
AbstractLeast Squares Support Vector Machine (LS-SVM) is powerful to solve problems such as multi-class nonlinear classification. In this study, a LS-SVM with kernel-based was applied to multi-class IC packaging type dataset classification problem. In the first stage, the greedy search algorithm in feature selection that reduced 15 features to 9 features was used to improve the LS-SVM. Then, the hyperparameters of LS-SVM were tuned by using 10-fold cross-validation procedure and a grid search mechanism. This study compared the classification performance of two classifiers, namely the LS-SVM with RBF kernel, LS-SVM with polynomial kernel and NN method, in 63 classes of IC packaging type dataset in the full model and feature redundant model. The results showed that, for the classification problem of multi-class IC packaging type dataset, in the full model and reduced model, the classification performance of LS-SVM with RBF kernel is better than that of LS-SVM with polynomial kernel and NN classifier. The ac...
Highlights
$EVWUDFW Least Squares Support Vector Machine (LS-Support Vector Machines (SVMs)) is powerful to solve problems such as multi-class nonlinear classification
The results showed that, for the classification problem of multi-class IC packaging type dataset, in the full model and reduced model, the classification performance of Least Squares Support Vector Machine (LS-SVM) with RBF kernel is better than that of LS-SVM with polynomial kernel and NN classifier
Regarding the multi-class IC packaging type classification problem, the classifier using the LS-SVM with RBF of the greedy feature reduced model has the highest classification performance in reducing the dataset complexity, with an accuracy rate at 85.29%. .H\ZRUGV: IC Packaging Type, Feature selection, Least Squares Support Vector Machine, Kernel-Based, Classification Accuracy
Summary
The main purposes of IC packaging include: 1) electric transmission, 2) signal transmission, 3) heat removal, 4) circuit protection. This paper demonstrated a neural network classifier model based on rough set feature selection in the multiclass IC package products classify. Hung 2 applied Variable Precision Rough Set (VPRS) and Backpropagation Neural network (BPNN) to establish the prediction models and classification criteria for the 5 product families. Except for TFBGA and LGA that have conditional features hard to recognize, the test accuracy rates of the other 3 product families (PBGA, QFP and FCBGA) were100%. Hsieh 9 added a feature to discuss whether the packaging manufacturing process uses the Ball Grid to solve the problem of difficulty in recognizing the conditional features in case of TFBGA and LGA. The results showed that the classification results based on approximate set could reach 100% after adding the characteristic feature of Ball Grid to correctly distinguish 5 product families
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: International Journal of Computational Intelligence Systems
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
