Ensemble Models of Learning Vector Quantization Based on Bootstrap Resampling

The purpose of this study is to improve the classification accuracy and stability of learning vector quantization using ensemble learning. We focused on an ensemble learning algorithm based on bootstrap resampling; this algorithm has been widely used in recent years. LVQs were extended to the ensemble model using three similar approaches: bagging, random forest, and double bagging. Through computational experiments using benchmark data, we investigated the compatibility between each approach and LVQ. The results showed that the double bagging approach was superior in ensemble LVQ.