Citation: Zhao, Y.; Song, A.; Qin, C.
Application of Noise Detection Using
Confidence Learning in Lightweight
Expression Recognition System. Appl.
Sci. 2022, 12, 4808. https://doi.org/
10.3390/app12104808
Academic Editor: Yosoon Choi
Received: 28 March 2022
Accepted: 5 May 2022
Published: 10 May 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Article
Application of Noise Detection Using Confidence Learning in
Lightweight Expression Recognition System
Yu Zhao
1,2,3
, Aiguo Song
1,2,3,
* and Chaolong Qin
1,2,3
1
The State Key Laboratory of Bioelectronics, Southeast University, Nanjing 210096, China;
220193286@seu.edu.cn (Y.Z.); 230198306@seu.edu.cn (C.Q.)
2
Jiangsu Key Lab of Remote Measurement and Control, Southeast University, Nanjing 210096, China
3
School of Instrument Science and Engineering, Southeast University, Nanjing 210096, China
* Correspondence: a.g.song@seu.edu.cn
Abstract:
Facial expression is an important carrier to reflect psychological emotion, and the lightweight
expression recognition system with small-scale and high transportability is the basis of emotional
interaction technology of intelligent robots. With the rapid development of deep learning, fine-
grained expression classification based on the convolutional neural network has strong data-driven
properties, and the quality of data has an important impact on the performance of the model. To solve
the problem that the model has a strong dependence on the training dataset and weak generalization
performance in real environments in a lightweight expression recognition system, an application
method of confidence learning is proposed. The method modifies self-confidence and introduces two
hyper-parameters to adjust the noise of the facial expression datasets.
A lightweight
model structure
combining a deep separation convolution network and attention mechanism is adopted for noise
detection and expression recognition. The effectiveness of dynamic noise detection is verified on
datasets with different noise ratios. Optimization and model training is carried out on four public
expression datasets, and the accuracy is improved by 4.41% on average in multiple test sample sets. A
lightweight expression recognition system is developed, and the accuracy is significantly improved,
which verifies the effectiveness of the application method.
Keywords:
confidence learning; expression recognition system; dynamic noise adjustment; datasets
optimization
1. Introduction
The expectation of a convenient daily life, an aging society, child care in two-job
families, and the shortage of nursing staff and other issues have created a large demand for
intelligent service robots. Due to the need to directly provide corresponding services for
users, the human–computer interaction of intelligent service robots needs to be comprehen-
sive and natural. Service robots need to recognize users’ emotions and respond in many
application scenarios which are defined as emotional interaction.
Facial expression is an important way for humans to express emotions and accounts
for 55% of the information in communication [
1
]. Facial expression recognition is widely
used in medical [
2
], business, teaching [
3
], criminal investigation, and other fields to
analyze people’s emotional states by capturing facial expressions.Therefore, the emotional
interaction intelligent robot based on expression recognition has become a research hotspot.
Expression recognition has been gradually studied by many subjects such as computer
science, psychology, cognitive science, neural computing, and so on since the 1990s. Two
core steps of expression recognition are feature extraction and expression classification.
Feature recognition is particularly important in traditional methods. Global feature extrac-
tion methods and local feature extraction methods represented by geometric and texture
feature extraction have been proposed. Machine learning is first applied to the expression
Appl. Sci. 2022, 12, 4808. https://doi.org/10.3390/app12104808 https://www.mdpi.com/journal/applsci
