Guixiang Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Guixiang Wang has a total of 56 co-authors. Here are the top co-authors by number of shared publications:
- Hai Liu
Central China Normal University
3 shared publications - Hui Li
School of Electronic Information and Communications, Huazhong University of Science and Technology, China
2 shared publications - Jiawen Li
School of Computer Science, Wuhan Donghu University, China
2 shared publications - Jinhang Lu
School of Computer Science, Wuhan Donghu University, Wuhan, Hubei, China
2 shared publications - Le He
School of Computer Science, Wuhan Donghu College, Wuhan, Hubei, China
2 shared publications - Li Zhao
National Engineering Research Center for E-learning, Central China Normal University, China
2 shared publications
Latest Publications
- Open Access
- 20 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 617-622
PleFER: Exploiting Plexiform Relationship Representation via Textural Architectonics Modeling for Facial Expression Recognition
Facial expression recognition (FER) for distinguishing emotional states is challenging due to several issues, including occlusions, extreme lighting conditions, and arbitrary face orientations. To address these challenges, we propose PleFER, a novel framework that leverages plexiform relationship and textural architectonics representations. This model introduces the plexiform relationship mining (PRM), textural architectonics calibration (TAC), and adaptive feature synthesis (AFS) modules. To the best of our knowledge, this study is the first to introduce plexiform relationships for mining architectonics representations in FER tasks. Experimental results on the RAF-DB, KDEF, and RaFD datasets demonstrate that the proposed PleFER model consistently outperforms state-of-the-art methods.
- 20 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 617-622
Open AccessPleFER: Exploiting Plexiform Relationship Representation via Textural Architectonics Modeling for Facial Expression Recognition
Facial expression recognition (FER) for distinguishing emotional states is challenging due to several issues, including occlusions, extreme lighting conditions, and arbitrary face orientations. To address these challenges, we propose PleFER, a novel framework that leverages plexiform relationship and textural architectonics representations. This model introduces the plexiform relationship mining (PRM), textural architectonics calibration (TAC), and adaptive feature synthesis (AFS) modules. To the best of our knowledge, this study is the first to introduce plexiform relationships for mining architectonics representations in FER tasks. Experimental results on the RAF-DB, KDEF, and RaFD datasets demonstrate that the proposed PleFER model consistently outperforms state-of-the-art methods.
- Open Access
- 20 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 54-59
Research on a Multimodal Physiological Feature Recognition Model for Anxiety Disorders Based on Random Forest
- 20 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 54-59
Open AccessResearch on a Multimodal Physiological Feature Recognition Model for Anxiety Disorders Based on Random Forest
- Open Access
- 20 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 1612-1617
Prediction of Existential Loneliness and Analysis of Influencing Factors Among Adolescents Based on Machine Learning
As a key dimension affecting adolescents' mental health, existential loneliness exhibits significant heterogeneity and dynamic developmental characteristics. Its accurate prediction and analysis of influencing factors are crucial for safeguarding adolescents' mental health. This study integrated four machine learning algorithms—logistic regression, random forest, support vector machine, and gradient boosting decision tree—to construct a classification prediction model for adolescents' existential loneliness, and systematically explored the core influencing factors and their interaction effects. The results showed that the gradient boosting decision tree model performed optimally, with a cross-sectional prediction accuracy of 89.7% and an AUC value of 0.943, which was significantly superior to traditional statistical methods. The model identified four core predictive factors: self-concept clarity, family support, friend support, and class identification integration, with a cumulative contribution rate of 86.6%. Moreover, there were significant interaction effects among these core factors—high-level synergy between self-concept clarity and family support could increase the probability of low existential loneliness among adolescents to 92.4%, and class identification integration could enhance the protective effect of friend support. Longitudinal validation indicated that the model had a prediction accuracy of 83.2% for changes in existential loneliness status, enabling effective identification of high-risk transition groups. By leveraging machine learning, this study achieved accurate prediction of adolescents' existential loneliness, revealed its influencing mechanisms and factor interaction effects, and provided sufficient theoretical basis for the early screening and targeted intervention of adolescents' existential loneliness.
- 20 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 1612-1617
Open AccessPrediction of Existential Loneliness and Analysis of Influencing Factors Among Adolescents Based on Machine Learning
As a key dimension affecting adolescents' mental health, existential loneliness exhibits significant heterogeneity and dynamic developmental characteristics. Its accurate prediction and analysis of influencing factors are crucial for safeguarding adolescents' mental health. This study integrated four machine learning algorithms—logistic regression, random forest, support vector machine, and gradient boosting decision tree—to construct a classification prediction model for adolescents' existential loneliness, and systematically explored the core influencing factors and their interaction effects. The results showed that the gradient boosting decision tree model performed optimally, with a cross-sectional prediction accuracy of 89.7% and an AUC value of 0.943, which was significantly superior to traditional statistical methods. The model identified four core predictive factors: self-concept clarity, family support, friend support, and class identification integration, with a cumulative contribution rate of 86.6%. Moreover, there were significant interaction effects among these core factors—high-level synergy between self-concept clarity and family support could increase the probability of low existential loneliness among adolescents to 92.4%, and class identification integration could enhance the protective effect of friend support. Longitudinal validation indicated that the model had a prediction accuracy of 83.2% for changes in existential loneliness status, enabling effective identification of high-risk transition groups. By leveraging machine learning, this study achieved accurate prediction of adolescents' existential loneliness, revealed its influencing mechanisms and factor interaction effects, and provided sufficient theoretical basis for the early screening and targeted intervention of adolescents' existential loneliness.
- Open Access
- 14 March 2025
- Published by Association for Computing Machinery (ACM)
- p. 1009-1013
A One-Stop Travel Service Platform Tailored for Wuhan University Students
- 14 March 2025
- Published by Association for Computing Machinery (ACM)
- p. 1009-1013
Open AccessA One-Stop Travel Service Platform Tailored for Wuhan University Students
- Open Access
- 14 March 2025
- Published by Association for Computing Machinery (ACM)
- p. 99-104
Segmented Image Retrieval Using MPEG-7 Similarity
- 14 March 2025
- Published by Association for Computing Machinery (ACM)
- p. 99-104
Open AccessSegmented Image Retrieval Using MPEG-7 Similarity
- Top Cited Papers
- 18 January 2025
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 104
Accurate identification of anxiety and depression based on the dlPFC in an emotional autobiographical memory task: A machine learning approach
- 18 January 2025
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 104
Top Cited PapersAccurate identification of anxiety and depression based on the dlPFC in an emotional autobiographical memory task: A machine learning approach
- Open Access
- 25 June 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (6) , e0299158
Physical activity on the mental health of children and adolescents during COVID-19 pandemic-induced school closures—A systematic review
Propose: To review published Physical Activity (PA) on the Mental Health of Children and Adolescents aged 5 to 18 years during COVID-19 pandemic-induced school closures. Methods: From the database creation to April 2022, 10 databases are retrieved, with 4427 records filtered, 14 included in this research. The research takes Agency for Healthcare Research and Quality (AHRQ) evaluation standards. Results: The thesis selects 14 studies from 6 countries, involving 400009 children and adolescents. These studies happened during the lockdown of COVID-19 (from December 2019 to April 2021). During the lockdown of COVID-19, schools were closed, which was considered part of a more extensive lockdown. Schools were closed for 1 to 4 weeks. There were 10 high quality studies (71.4%) and 4 medium quality studies (28.6%). 4 studies report that the pandemic reduces the time of PA but increases the time of watching screen and sitting. 10 studies (71.4%) identify that PA is positive for the mental health, because it helps reduce mental symptoms to a certain extent, especially anxiety, depression, and emotional disorders. 5 studies show that PA may not improve the mental health of children and adolescents under 12 during the pandemic. 4 studies indicate that the influence of PA on mental health of children and adolescents is determined by the amount of activity, including the extent, intensity, frequency, and duration, etc. Conclusions: In this narrative synthesis of reports from the class suspension period, reports that PA has a improve on the mental health of children and adolescents to a certain extent. it is found that PA may be helpful in reducing mental health symptoms of children and adolescents who are influenced by class suspension because of the COVID-19 pandemic. Therefore, stakeholders of the mental health of children and adolescents around the world should recommend PA because it is a practicable and beneficial way for long-term mental support.
- 25 June 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (6) , e0299158
Open AccessPhysical activity on the mental health of children and adolescents during COVID-19 pandemic-induced school closures—A systematic review
Propose: To review published Physical Activity (PA) on the Mental Health of Children and Adolescents aged 5 to 18 years during COVID-19 pandemic-induced school closures. Methods: From the database creation to April 2022, 10 databases are retrieved, with 4427 records filtered, 14 included in this research. The research takes Agency for Healthcare Research and Quality (AHRQ) evaluation standards. Results: The thesis selects 14 studies from 6 countries, involving 400009 children and adolescents. These studies happened during the lockdown of COVID-19 (from December 2019 to April 2021). During the lockdown of COVID-19, schools were closed, which was considered part of a more extensive lockdown. Schools were closed for 1 to 4 weeks. There were 10 high quality studies (71.4%) and 4 medium quality studies (28.6%). 4 studies report that the pandemic reduces the time of PA but increases the time of watching screen and sitting. 10 studies (71.4%) identify that PA is positive for the mental health, because it helps reduce mental symptoms to a certain extent, especially anxiety, depression, and emotional disorders. 5 studies show that PA may not improve the mental health of children and adolescents under 12 during the pandemic. 4 studies indicate that the influence of PA on mental health of children and adolescents is determined by the amount of activity, including the extent, intensity, frequency, and duration, etc. Conclusions: In this narrative synthesis of reports from the class suspension period, reports that PA has a improve on the mental health of children and adolescents to a certain extent. it is found that PA may be helpful in reducing mental health symptoms of children and adolescents who are influenced by class suspension because of the COVID-19 pandemic. Therefore, stakeholders of the mental health of children and adolescents around the world should recommend PA because it is a practicable and beneficial way for long-term mental support.
- Open Access
- 26 April 2024
- Published by Association for Computing Machinery (ACM)
- p. 1213-1218
Research on depression diagnosis based on the Machine Learning
Depression is a common mental disorder that has serious effects on patients' physical and mental health. Diagnosis of depression in clinical practice mainly relies on the experience and professional knowledge of physicians, which results in inaccurate diagnosis, misdiagnosis, and missed diagnosis. In recent years, the development of machine learning technology has provided new possibilities for automated diagnosis of depression. This paper proposes a machine learning-based diagnosis model for depression and validates the model through experiments. The model uses the deep neural network gcForest (multi-granularity cascade forest) to classify patients with depression and normal individuals, and uses multiple physiological features as input, including electroencephalogram data, heart rate variability, skin conductance activity, etc. The experimental results show that the model can effectively distinguish patients with depression from normal individuals and has high accuracy, precision, recall, and F1 scores. The machine learning-based diagnosis model for depression proposed in this study has certain clinical application value, can provide doctors with diagnostic assistance tools, and is expected to provide more effective means for early warning and treatment of depression in the future.
- 26 April 2024
- Published by Association for Computing Machinery (ACM)
- p. 1213-1218
Open AccessResearch on depression diagnosis based on the Machine Learning
Depression is a common mental disorder that has serious effects on patients' physical and mental health. Diagnosis of depression in clinical practice mainly relies on the experience and professional knowledge of physicians, which results in inaccurate diagnosis, misdiagnosis, and missed diagnosis. In recent years, the development of machine learning technology has provided new possibilities for automated diagnosis of depression. This paper proposes a machine learning-based diagnosis model for depression and validates the model through experiments. The model uses the deep neural network gcForest (multi-granularity cascade forest) to classify patients with depression and normal individuals, and uses multiple physiological features as input, including electroencephalogram data, heart rate variability, skin conductance activity, etc. The experimental results show that the model can effectively distinguish patients with depression from normal individuals and has high accuracy, precision, recall, and F1 scores. The machine learning-based diagnosis model for depression proposed in this study has certain clinical application value, can provide doctors with diagnostic assistance tools, and is expected to provide more effective means for early warning and treatment of depression in the future.
- Open Access
- 26 April 2024
- Published by Association for Computing Machinery (ACM)
- p. 707-712
Blind infrared spectral signal reconstruction with improved total variation constraint
In the infrared spectral domain, the presence of random noise and the overlapping of adjacent peaks pose significant challenges. To tackle these issues, we have introduced a novel model for the reconstruction of infrared spectra, incorporating enhancements to the total variation regularization. The maximum a posteriori (MAP) theory is introduced to construct the signal reconstruction model for latent infrared spectrum. The proposed model can remove the random spectral noises and produce the new peaks. In the MAP architecture, the prior probability is modeled as the improved total variation constraint, the Gaussian distribution-based spectral noise is constructed for the likelihood probability distribution. Then, the proposed infrared spectral signal model is optimized by the famous split Bregman iteration approach. The high-quality infrared spectrum and instrument function can be estimated simulated by the proposed model on the simulated and real spectral signals. All the comparative experiments were conducted using Matlab 2022, illustrating the effective performance in suppressing spectral noises and splitting overlap peaks.
- 26 April 2024
- Published by Association for Computing Machinery (ACM)
- p. 707-712
Open AccessBlind infrared spectral signal reconstruction with improved total variation constraint
In the infrared spectral domain, the presence of random noise and the overlapping of adjacent peaks pose significant challenges. To tackle these issues, we have introduced a novel model for the reconstruction of infrared spectra, incorporating enhancements to the total variation regularization. The maximum a posteriori (MAP) theory is introduced to construct the signal reconstruction model for latent infrared spectrum. The proposed model can remove the random spectral noises and produce the new peaks. In the MAP architecture, the prior probability is modeled as the improved total variation constraint, the Gaussian distribution-based spectral noise is constructed for the likelihood probability distribution. Then, the proposed infrared spectral signal model is optimized by the famous split Bregman iteration approach. The high-quality infrared spectrum and instrument function can be estimated simulated by the proposed model on the simulated and real spectral signals. All the comparative experiments were conducted using Matlab 2022, illustrating the effective performance in suppressing spectral noises and splitting overlap peaks.
- Open Access
- 26 April 2024
- Published by Association for Computing Machinery (ACM)
- p. 605-609
Optimizing Person Re-Identification: Exploring Deep Learning Methods Based on Attention Mechanism-driven Weighted Feature Pyramid Network
China has established the world's largest video surveillance system, generating massive amounts of video data every day. However, relying solely on manual methods for analyzing these data is inefficient. Pedestrian re-identification technology can search for specific pedestrians in a large number of video sequences or images across scenes and cameras, playing an important role in public safety, smart security, and other fields. However, factors such as occlusion, low resolution, and lighting in real-world scenarios lead to low pedestrian re-identification rates. To address this issue, this study focuses on the field of deep learning for pedestrian re-identification and proposes a method based on attention mechanism and weighted feature pyramid network. This method effectively extracts and represents pedestrian image features by introducing attention mechanism and weighted feature fusion. We designed a weighted feature pyramid network structure, achieving fusion of features at different scales, thereby improving the network's generalization ability and robustness. Additionally, we employed channel attention and spatial attention mechanisms to precisely focus on important information, optimizing the feature extraction process. Experimental validation demonstrates that this method significantly enhances the performance and accuracy of pedestrian re-identification models, providing important technical support and methodological guidance for further advancement in this field.
- 26 April 2024
- Published by Association for Computing Machinery (ACM)
- p. 605-609
Open AccessOptimizing Person Re-Identification: Exploring Deep Learning Methods Based on Attention Mechanism-driven Weighted Feature Pyramid Network
China has established the world's largest video surveillance system, generating massive amounts of video data every day. However, relying solely on manual methods for analyzing these data is inefficient. Pedestrian re-identification technology can search for specific pedestrians in a large number of video sequences or images across scenes and cameras, playing an important role in public safety, smart security, and other fields. However, factors such as occlusion, low resolution, and lighting in real-world scenarios lead to low pedestrian re-identification rates. To address this issue, this study focuses on the field of deep learning for pedestrian re-identification and proposes a method based on attention mechanism and weighted feature pyramid network. This method effectively extracts and represents pedestrian image features by introducing attention mechanism and weighted feature fusion. We designed a weighted feature pyramid network structure, achieving fusion of features at different scales, thereby improving the network's generalization ability and robustness. Additionally, we employed channel attention and spatial attention mechanisms to precisely focus on important information, optimizing the feature extraction process. Experimental validation demonstrates that this method significantly enhances the performance and accuracy of pedestrian re-identification models, providing important technical support and methodological guidance for further advancement in this field.