Cost-sensitive collaborative representation based classification via probability estimation with addressing the class imbalance

摘要

Collaborative representation has been successfully used in pattern recognition and machine learning. However, most existing collaborative representation classification methods are to achieve the highest classification accuracy, assuming the same losses for different misclassifications. This assumption, however, may not hold in many real-word applications as different types of misclassification could lead to different losses. Meanwhile, the class distribution of data is highly imbalanced in real-world applications. To address this problem, a novel Cost-Sensitive Collaborative Representation based Classification (CSCRC) method via Probability Estimation with Addressing the Class Imbalance was proposed. Unlike traditional methods, the class label of test samples is predicted by minimizing the misclassification losses which are obtained via computing the posterior probabilities. In this paper, a Gaussian function was defined as a probability distribution of collaborative representation coefficient vector and the probability distribution was transformed into collaborative representation framework via logarithmic operator. The experiments show that our proposed method performs competitively compared with existing methods.