One-class classification for spontaneous facial expression analysis

In this paper, we explore one-class classification application in recognizing emotional and nonemotional facial expressions occurred in a realistic human conversation setting - adult attachment interview (AAI). Although emotional facial expressions are defined in terms of facial action units in the psychological study, non-emotional facial expressions have not distinct description. It is difficult and expensive to model non-emotional facial expressions. Thus, we treat this facial expression recognition as a one-class classification problem which is to describe target objects (i.e. emotional facial expressions) and distinguish them from outliers (i. e. non-emotional ones). We first apply Kernel whitening to map the emotional data in a kernel subspace with unit variances in all directions. Then, we use support vector data description (SVDD) for the classification which is to directly fit a boundary with minimal volume around the target data. We present our preliminary experiments on the AAI data, and compare Kernel whitening SVDD with PCA+SVDD and PCA+Gaussian methods