Article
Chemistry, Multidisciplinary
Yi Zhu, Xinke Zhou, Xindong Wu
Summary: Unsupervised domain adaptation involves transferring knowledge from a labeled source domain to unlabeled target domains for target learning tasks. This paper proposes an unsupervised domain adaptation method based on a stacked convolutional sparse autoencoder, which performs layer projection to obtain higher-level representations. The method addresses training problems and performance degradation issues in feature learning.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Artificial Intelligence
Souvik Phadikar, Nidul Sinha, Rajdeep Ghosh
Summary: This study proposes an innovative method to transform EEG signals into weight vectors of an autoencoder to address the issue of informative confusion in multiclass classification of MI based on EEG data. The extracted features from the weight vectors are used in a support vector machine as a classifier network to improve the decoding performance of BCIs.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Jianping Gou, Nannan Xie, Jinhua Liu, Baosheng Yu, Weihua Ou, Zhang Yi, Wu Chen
Summary: This article introduces a hierarchical graph augmented stacked autoencoder (HGSAE) method for unsupervised multi-view representation learning. The method preserves the geometric information of multi-view data through local and non-local graph regularizations and learns a general representation by reconstructing each single view. Extensive experiments demonstrate the effectiveness of the proposed method in unsupervised representation learning.
INFORMATION FUSION
(2024)
Article
Engineering, Electrical & Electronic
Sanjay Roka, Manoj Diwakar
Summary: Due to the rise in crime and terrorism, security concerns are increasing daily. Surveillance cameras are now an essential tool for detecting abnormal behavior, but most existing systems have low performance and accuracy due to noise in the videos. To address this issue, this study proposes a method for denoising images and detecting abnormal activity using a deep learning model. The results demonstrate superior accuracy and performance compared to existing methods in benchmark datasets. Overall, this research has significant importance in improving security surveillance systems. Evaluation: 7/10.
JOURNAL OF ELECTRONIC IMAGING
(2023)
Article
Computer Science, Artificial Intelligence
Trishita Dhara, Pawan Kumar Singh, Mufti Mahmud
Summary: This paper proposes a fuzzy ensemble-based deep learning approach for emotion recognition from EEG signals. The proposed model achieves high accuracies on benchmark datasets and is considered robust for emotion recognition.
COGNITIVE COMPUTATION
(2023)
Article
Computer Science, Interdisciplinary Applications
Ilkay Yildiz, Rachael Garner, Matthew Lai, Dominique Duncan
Summary: This article introduces an unsupervised seizure identification method based on deep learning. The method uses a variational autoencoder (VAE) to train on raw EEG and identifies seizures based on reconstruction errors. The experimental results show that our method can successfully distinguish seizures from non-seizure activity.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2022)
Article
Computer Science, Information Systems
Yi Zhu, Xindong Wu, Jipeng Qiang, Yunhao Yuan, Yun Li
Summary: The purpose of unsupervised domain adaptation is to leverage knowledge from a source domain with a different data distribution from the target domain to improve learning in the target domain. Deep learning methods based on autoencoders have achieved good performance in representation learning, but existing methods often fail to find the real cross-domain features. To address this problem, we propose a novel representation learning method called IAUDA, which combines different autoencoders and introduces a sparse autoencoder for feature compression. Experimental results show that our proposed method is more effective than several state-of-the-art baseline methods.
FRONTIERS OF COMPUTER SCIENCE
(2023)
Article
Computer Science, Information Systems
Mohammad Asif, Sudhakar Mishra, Majithia Tejas Vinodbhai, Uma Shanker Tiwary
Summary: Emotion recognition using EEG signals is a promising field in Brain-Computer Interfaces. To overcome the limitations of existing emotion databases, we designed an experiment where participants freely reported their emotional feelings while watching emotional stimuli. Our dataset, DENS, showed higher accuracy in classifying emotional events compared to benchmark datasets DEAP and SEED.
Article
Computer Science, Artificial Intelligence
Fan Xu, Lei Wang
Summary: An enhanced stacked autoencoder based on an exponent weight moving average in a deep learning model is proposed to eliminate noise from the health indicator and reduce dependence on manual experience. Experimental results demonstrate that the extracted health indicator curve is smoother and has higher monotonicity compared to other models.
ADVANCED ENGINEERING INFORMATICS
(2022)
Article
Computer Science, Artificial Intelligence
Mamadou Dian Bah, Adel Hafiane, Raphael Canals
Summary: Crop row detection is crucial for smart farming, and deep learning approaches have been widely studied for this task. However, the requirement of large labeled datasets poses a challenge, especially in agriculture where labeling data is tedious and expensive. Graph-based unsupervised techniques offer a promising alternative by incorporating structured information such as plant relationships.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Wanzeng Kong, Xulin Song, Junfeng Sun
Summary: This paper introduces a novel feature extraction method utilizing phase synchronization for emotion recognition based on EEG. By estimating EEG phase synchronization indexes, performing principal component analysis, and utilizing Sparse Representation based Classification, the proposed method achieves superior performance in emotion classification.
MULTIMEDIA TOOLS AND APPLICATIONS
(2021)
Article
Biology
Yi Wei, Yu Liu, Chang Li, Juan Cheng, Rencheng Song, Xun Chen
Summary: Deep learning has made remarkable progress in emotion recognition based on Electroencephalogram (EEG), with convolutional neural networks (CNNs) being the most commonly used models. However, due to their local feature learning mechanism, CNNs struggle to capture global contextual information. This paper proposes a Transformer Capsule Network (TC-Net) that overcomes this limitation and achieves state-of-the-art performance in EEG-based emotion recognition.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Computer Science, Artificial Intelligence
Han Zhao, Xu Yang, Cheng Deng, Dacheng Tao
Summary: In this study, we propose a structure-adaptive graph contrastive learning framework to capture potential discriminative relationships for improved graph representation learning.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Engineering, Multidisciplinary
Raja Majid Mehmood, Muhammad Bilal, S. Vimal, Seong-Whan Lee
Summary: By using a novel Hjorth-feature-based emotion recognition model, this study explores a wider set of emotion classes and achieves better accuracy in EEG-based emotion recognition.
Article
Engineering, Biomedical
Haoming Zhang, Mingqi Zhao, Chen Wei, Dante Mantini, Zherui Li, Quanying Liu
Summary: The study introduces a benchmark EEG dataset EEGdenoiseNet for training and testing DL-based denoising models, and evaluates the denoising performance of four classical networks, indicating the great potential of DL methods under high noise contamination.
JOURNAL OF NEURAL ENGINEERING
(2021)
Review
Chemistry, Analytical
Junhai Luo, Ying Han, Liying Fan
Review
Chemistry, Analytical
Junhai Luo, Liying Fan, Shan Wu, Xueting Yan
Article
Chemistry, Analytical
Junhai Luo, Xiaoting He
Article
Engineering, Electrical & Electronic
Junhai Luo, Ying Han
Article
Multidisciplinary Sciences
Junhai Luo, Lei Ye
SCIENTIFIC REPORTS
(2019)
Article
Engineering, Electrical & Electronic
Junhai Luo, Ying Han, Xiaoting He
Article
Engineering, Electrical & Electronic
Junhai Luo, Xiaoting He
IET COMMUNICATIONS
(2020)
Article
Chemistry, Analytical
Junhai Luo, Yang Yang, Zhiyan Wang, Yanping Chen, Man Wu
Article
Chemistry, Analytical
Junhai Luo, Zhiyan Wang, Yanping Chen, Man Wu, Yang Yang
Review
Computer Science, Information Systems
Junhai Luo, Yang Yang, Zhiyan Wang, Yanping Chen
Summary: Underwater localization, a significant component of ocean exploration, has attracted extensive attention in both military and civil fields. Underwater wireless sensor networks (UWSNs) are favored for their low cost and convenience, strengthening the trinity of land, sea, and air as an important part of the Internet of Things (IoT). Many scholars have optimized localization algorithms and introduced new methods to better locate target nodes, promoting the development of related fields.
IEEE INTERNET OF THINGS JOURNAL
(2021)
Article
Computer Science, Theory & Methods
Junhai Luo, Zhiyan Wang, Ming Xia, Linyong Wu, Yuxin Tian, Yu Chen
Summary: Path planning is crucial for the flexible deployment and performance of unmanned aerial vehicles communication networks (UAVCN). This article provides a comprehensive review of UAVCN path planning, including network structure and performance evaluation, generic UAV path planning algorithms, and path planning algorithms specifically designed for UAVCN. The advantages, disadvantages, functional problems, challenges, solutions, state-of-the-art, and representative results of each path planning algorithm are discussed. Furthermore, future research directions for UAVCN path planning are proposed to assist researchers.
ACM COMPUTING SURVEYS
(2023)
Article
Environmental Sciences
Siying Cao, Jiakun Deng, Junhai Luo, Zhi Li, Junsong Hu, Zhenming Peng
Summary: This study proposes a robust scheme for automatically detecting infrared small targets, which improves the accuracy of detecting dim and small targets in complex scenes. It has competitive performance with state-of-the-art algorithms and low time consumption, making it beneficial for practical applications.
Article
Geochemistry & Geophysics
Fengyi Wu, Hang Yu, Anran Liu, Junhai Luo, Zhenming Peng
Summary: Infrared small target detection is crucial for both civil and military applications, but current methods face challenges in dealing with complex scenes, distinguishing targets from similar objects, and utilizing temporal information effectively. To overcome these limitations, we propose an innovative approach that takes advantage of the spatiotemporal structure of infrared images. By constructing a 4-D infrared tensor and decomposing it into lower dimensional tensors using the tensor train and tensor ring techniques, we formulate the ISTD problem as a sparse plus low-rank decomposition problem. We validate our approach on multiple datasets and compare it with state-of-the-art techniques in terms of detection accuracy and background suppression, demonstrating its superiority.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Computer Science, Information Systems
Junhai Luo, Yanping Chen, Man Wu, Yang Yang
Summary: Underwater wireless sensor networks are a hot research field with various limitations and challenges, and well-designed routing protocols can effectively address these issues. Existing underwater routing protocols can be classified into three categories: energy-based, data-based, and geographic information-based. Research challenges and future directions in underwater routing protocols are worth exploring.
IEEE COMMUNICATIONS SURVEYS AND TUTORIALS
(2021)