Improvement of Feature Localization for Facial Expressions by Adding Noise

Abstract

This paper investigates feature localization abilities upon injecting noise into the convolutional neural network (CNN). The proposed model intended to classify the 7 human emotional states based on facial expressions and it is shown to perform better than the earlier convolutional neural network. The internal representation of learned features emerges and a more accurate localization of those features appears when independent Gaussian noises are added to certain joints during the deep network training. We observed that the weights after the noise contaminated units lead to output that is more definite. Such behavior improves the network generalization through automatic structuration. We confirmed this by emotion classification experiments on KDEF black and Cohen-Kanade + datasets based on facial expression.

Transformer-based Siamese and Triplet Networks for Facial Expression Intensity Estimation

“…We experimented with multiple dropout placements and the latter placement assisted the convergence, namely the third and fourth transformer blocks. The benefits of the latter placement of regularization in neural networks are aligned with the work of Sabri and Kurita [73,74].…”