An efficient linear regression classifier
Hai Wang, Fei Hao · 2012
Pattern recognition is one of the most important research topics in recent days. In this area, one of the crucial problems is the design of the classifier. The most classic and simplest classifier is the K-NN algorithm, and it has been widely used in many fields such as text recognition and face recognition. In this paper, we propose an efficient and simple classifier, called linear regression classifier (LRC), which considers the nature of the different patterns. We first propose LRC-LSE algorithm based on the LSE estimation algorithm, and classify the data according to the linear regression errors. In addition, considering the multi-collinearity, we propose LRC-PLS algorithm based on the PLS estimation approach, further, we evaluate our algorithm in face recognition. Experimental results demonstrate that our algorithm achieves the better classification results than K-NN algorithm with a lower computational cost.