Building a Compact MQDF Classifier by Sparse Coding and Vector Quantization Technique

Abstract

The modified quadratic discriminant function (MQDF) is a very popular handwritten Chinese character classifier thanks to its high performance with low computational complexity. However, it suffers from high memory requirement for the storage of its parameters. This paper proposes a compact MQDF classifier developed by integrating sparse coding and vector quantization (VQ) technique. To be specific, we use sparse coding to represent the parameters of MQDF in sparsity first, and then employ the VQ technique to further compress the sparse coding. The proposed method is evaluated by comparing the performance with three models, i.e., the original MQDF classifier, the compact MQDF classifier using the VQ technique, and the compact MQDF classifier using sparse coding. The effectiveness of our proposed approach has been confirmed and demonstrated by comparative experiments on ICDAR2013 competition dataset.