LAUSR

An important challenge for facial expression recognition (FER) is that real-world training data are usually imbalanced. Although many deep learning approaches have been proposed to enhance the discriminative power of deep expression features and enable a good predictive effect, few works have focused on the multiclass imbalance problem. When supervised by softmax loss (SL), which is widely used in FER, the classifier is often biased against minority categories (i.e., smaller interclass angular distances). In this letter, we present advanced softmax loss (ASL) to mitigate the bias induced by data imbalance and hence increase accuracy and reliability. The proposed ASL essentially magnifies the interclass diversity in the angular space to enhance discriminative power in every category. The proposed loss can easily be implemented in any deep network. Extensive experiments on the FER2013 and real-world affective faces (RAF) databases demonstrate that ASL is significantly more accurate and reliable than many state-of-the-art approaches and that it can easily be plugged into other methods and improves their performance.