Article
Environmental Sciences
Tao Liu, Lexie Yang, Dalton Lunga
Summary: This study introduces a novel change detection method that combines deep learning and OBIA techniques, achieving higher accuracy in multi-class change detection tasks. Additionally, it compares three common feature fusion schemes for change detection accuracy for the first time.
REMOTE SENSING OF ENVIRONMENT
(2021)
Article
Chemistry, Analytical
Junjie Wang, Li Bi, Pengxiang Sun, Xiaogang Jiao, Xunde Ma, Xinyi Lei, Yongbin Luo
Summary: In this paper, a deep-learning-based defect detection method for photovoltaic cells is proposed, which addresses the challenges of scarce data and data imbalance. The method includes data enhancement, category weight assignment, and feature fusion using ResNet152-Xception. Experimental results show that the proposed method achieves high accuracy in PV cell defect detection.
Article
Computer Science, Hardware & Architecture
Juan-juan Fu, Xing-lan Zhang
Summary: This paper proposes a feature fusion technique based on gradient importance enhancement to improve the accuracy and generalization ability of the intrusion detection model in the current unstable network security situation.
Article
Geography, Physical
Pan Chen, Bing Zhang, Danfeng Hong, Zhengchao Chen, Xuan Yang, Baipeng Li
Summary: This paper introduces a feature-constrained change detection network (FCCDN) that utilizes deep learning techniques and constrains features in both bitemporal feature extraction and feature fusion. By building a dual encoder-decoder network backbone and a nonlocal feature pyramid network, as well as a dense connection-based feature fusion module, the network achieves state-of-the-art performance on the change detection task. Moreover, accurate bitemporal semantic segmentation results are achieved for the first time without using semantic segmentation labels, which is crucial for cost-saving in change detection applications.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2022)
Article
Computer Science, Artificial Intelligence
Federica Colonnese, Francesco Di Luzio, Antonello Rosato, Massimo Panella
Summary: This study introduces a new method for detecting ASD in children through gait analysis and deep learning. By extracting features from videos and combining gait analysis with deep learning, it provides a noninvasive and objective assessment of neurodevelopmental disorders.
INTERNATIONAL JOURNAL OF NEURAL SYSTEMS
(2023)
Article
Environmental Sciences
Qi Zhang, Yao Lu, Sicheng Shao, Li Shen, Fei Wang, Xuetao Zhang
Summary: Remote sensing change detection aims to identify changed pixels from two temporal images of the same location. Existing models use encoder-decoder structures and Siamese networks, but face challenges related to symmetry of change features, underutilization of the encoder parameters, and problems with sample balance and edge region detection. To address these issues, this paper proposes a mutual feature-aware network (MFNet) that includes a symmetric change feature fusion module, a mutual feature-aware module, and a loss function for edge regions. Experimental results demonstrate the effectiveness of MFNet, outperforming advanced algorithms with F1 scores of 83.11% and 91.52% on SYSU-CD and LEVIR-CD datasets, respectively.
Article
Engineering, Multidisciplinary
Xiaomeng Li, Yi Wang, Hulin Ruan, Dong Wang, Yi Qin, Baoping Tang
Summary: In this paper, a novel method based on deep transient feature learning is proposed to extract repetitive vibration transients and suppress in-band noise simultaneously. By constructing a deep model with simulated template signals and mapping noisy signal's time-frequency distribution (TFD) image to pure repetitive transients' TFD image, clean repetitive transients can be reconstructed effectively. Experimental validation results show the proposed method is reliable and effective for repetitive transients extracting.
Article
Computer Science, Artificial Intelligence
Jin Zhang, Qiuwei Liang, Qianqian Guo, Jinyu Yang, Qing Zhang, Yanjiao Shi
Summary: The article introduces a Residual Refinement Network (R(2)Net) method for salient object detection, which improves the performance of salient object detection through the fusion strategy of multi-scale features and contextual features. Experimental results demonstrate that the proposed method performs excellently on multiple benchmark datasets.
IMAGE AND VISION COMPUTING
(2022)
Article
Automation & Control Systems
Haifeng Wang, Lvjiyuan Jiang, Qian Zhao, Hao Li, Kai Yan, Yang Yang, Songlin Li, Yungang Zhang, Lianliu Qiao, Cuilian Fu, Hong Yin, Yun Hu, Haibin Yu
Summary: Deep learning-based target detection techniques have had a significant impact on daily life, with the feature pyramid being a widely utilized method for multiscale target detection. However, issues such as multiscale feature alignment and non-local feature fusion exist in the pyramid structure. The proposed progressive network structure effectively addresses these problems and shows improved performance compared to other state-of-art methods.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2021)
Article
Computer Science, Artificial Intelligence
Sunil Kumar Prabhakar, Dong-Ok Won
Summary: The diagnosis of cardiovascular diseases is crucial in medicine. Heart sound, which is influenced by blood turbulence and cardiac structures, plays a significant role in the early detection of heart diseases. Phonocardiogram (PCG) is a non-invasive technique used for heart sound analysis. This paper proposes efficient models for PCG signal classification, utilizing techniques like semi-supervised Non-negative Matrix Factorization (NMF), Brain Storming (BS) algorithm, Genetic Programming (GP), dimensionality reduction, and deep learning techniques. The experimental results demonstrate a high classification accuracy of 95.39% using the semi-supervised NMF concept with ABS-GP technique and Support Vector Machine (SVM) classifier.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Andrea Loddo, Giacomo Meloni, Barbara Pes
Summary: The study demonstrates the potential of artificial intelligence techniques in early diagnosis and detection of COVID-19 infections through high-dimensional data analysis, presenting the performance of various classification models.
Article
Computer Science, Artificial Intelligence
Kunfeng Wang, Yadong Wang, Shuqin Zhang, Yonglin Tian, Dazi Li
Summary: This article proposes a self-learning multi-scale object detection network, named SLMS-SSD, which balances the semantic information and spatial information to effectively improve the accuracy of object detection, especially for small object detection.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Chemistry, Multidisciplinary
Daniyal Alghazzawi, Omaimah Bamasag, Hayat Ullah, Muhammad Zubair Asghar
Summary: DDoS attacks pose a serious risk to computer networks and systems, but using machine learning/deep learning for detection can be helpful. Existing research has utilized ML classifiers and conventional methods to predict DDoS attacks, but accuracy remains a challenge.
APPLIED SCIENCES-BASEL
(2021)
Article
Computer Science, Artificial Intelligence
Hui-juan Zhu, Yang Li, Liang-min Wang, Victor S. Sheng
Summary: The continuous malware attacks on smartphones, especially on the dominant platform Android, pose a serious security threat to users. Data-driven methods based on machine learning algorithms show promise in defending against these attacks. This paper explores the limitations of such methods in improving malware detection performance and proposes a multi-model ensemble framework called MEFDroid, which combines individual predictors and utilizes hybrid deep learning based feature extraction methods to learn meaningful features from raw data.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Chao Dai, Chen Pan, Wei He
Summary: This paper proposes a feature extraction and fusion network (EFNet) that effectively integrates high-level semantic features and low-level image features, improving the performance of salient object detection.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Xin Lu, Yao Deng, Ting Sun, Yi Gao, Jun Feng, Xia Sun, Richard Sutcliffe
Summary: The research proposes a sentence matching method based on multi keyword-pair matching to represent the semantic relationship between sentences and avoid the interference of redundancy and noise. Experimental results show that this method can achieve state-of-the-art performance in several tasks.
APPLIED INTELLIGENCE
(2022)
Article
Biochemical Research Methods
Hansheng Li, Jianping Li, Yuxin Kang, Chunbao Wang, Feihong Liu, Wenli Hui, Qirong Bo, Lei Cui, Jun Feng, Lin Yang
Summary: Diagnostic pathology is crucial for identifying carcinomas, and accurate quantification of pathological images can provide objective clues. The Global Bank (GLB) pathway has been proposed to guide the extraction of more RoI features, significantly improving performance and increasing the accuracy of quantitative results.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2022)
Article
Mathematical & Computational Biology
Zhizezhang Gao, Yan Zhang, RuiPeng Zhang, Xia Sun, Jun Feng
Summary: Both traditional teaching and online teaching emphasize individualized education. However, the process of exploring improvements in instructional design is hindered by the challenging task of collecting data. Existing research primarily focuses on students' exam scores and overlooks their daily practice. In this study, we propose an experimental paradigm of programming performance analysis based on students' daily practice-exam records and collect a comprehensive time-series dataset, including students' individual attributes, learning behavior, and performance. We then analyze the time-series dataset using generalized estimating equations (GEE) to examine the impact of individual attributes and learning behavior on performance. This is the first application of GEE for ordinal multinomial responses in this research field, from which we conclude that gender and major do contribute to differences in programming learning. Longer answer times and shorter cost times are associated with better performance. Regardless of gender, students tend to cram for exams and perform slightly worse in daily exercises. Finally, we provide teaching mode decisions for universities based on two important individual attributes and recommend different teaching methods for students of different genders at different time points.
COMPUTATIONAL INTELLIGENCE AND NEUROSCIENCE
(2022)
Article
Computer Science, Information Systems
Xiaoxi Zhang, Yuan Gao, Xin Wang, Jun Feng, Yan Shi
Summary: This paper investigates the problem of transportation mode identification using GPS trajectories and geographic information, and proposes a geographic information-fused semi-supervised method. The proposed method can train an excellent transportation mode identification model with only a few labeled samples.
ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION
(2022)
Article
Computer Science, Artificial Intelligence
Ephrem Afele Retta, Eiad Almekhlafi, Richard Sutcliffe, Mustafa Mhamed, Haider Ali, Jun Feng
Summary: This article introduces the Amharic Speech Emotion Dataset (ASED) which consists of four dialects and five emotions. It is the first dataset for Speech Emotion Recognition (SER) in Amharic. The dataset was created by 65 native Amharic speakers who recorded 2,474 sound samples. The resulting dataset is freely available for download.
ACM TRANSACTIONS ON ASIAN AND LOW-RESOURCE LANGUAGE INFORMATION PROCESSING
(2023)
Article
Engineering, Civil
Boting Qu, Xinyu Ren, Jun Feng, Xin Wang
Summary: Ridesplitting is a convenient and budget-friendly for-hire transportation service that arranges shared rides on the fly, with effective rider allocation being a crucial component. The DRPP method proposed in this paper utilizes a grid network and historical GPS trajectories to predict pick-up probabilities and travel times, using ILSAS and TKdS-tree to improve efficiency for matching drivers and riders, which outperformed other methods in service rate, share rate, and rider waiting time in experiments.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Engineering, Biomedical
Feihong Liu, Junwei Yang, Mingyue Feng, Zhiming Cui, Xiaowei He, Luping Zhou, Jun Feng, Dinggang Shen
Summary: Phase correction is used to reconstruct real-valued diffusion MRI data by estimating the noise-free background phase. However, signal-loss and artifacts can still occur. In this paper, we propose a complex polar coordinate system (CPCS) to analyze the phase correction procedure and identify its limitations. Based on CPCS, we develop a quantitative criterion to better exploit the background phase and propose a phase calibration procedure to improve phase correction. Experimental results on synthetic and real diffusion MRI data demonstrate the effectiveness of our proposed method in reducing signal-loss and eliminating artifacts in FA maps.
COMPUTERIZED MEDICAL IMAGING AND GRAPHICS
(2023)
Article
Computer Science, Hardware & Architecture
Dekui Wang, Jun Feng, Wei Zhou, Xingxing Hao, Xiaodan Zhang
Summary: This article presents a fast FPGA connection router called FCRoute, which is based on a novel soft routing-space pruning algorithm. FCRoute classifies routing resource nodes into high-priority and low-priority ones and consists of a fast maze search and a backtracking process. By avoiding the exploration of the majority of low-priority nodes, FCRoute maintains runtime efficiency while ensuring global search ability.
IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Xia Sun, Bo Li, Richard Sutcliffe, Zhizezhang Gao, Wenying Kang, Jun Feng
Summary: Students can develop their skills by completing a series of tailored exercises, which is more effective than choosing exercises from online sources themselves. This paper presents a novel approach called Weighting-based Student Exercise Matrix Factorization (Wse-MF) that combines student learning ability and exercise difficulty. The research results demonstrate that Wse-MF outperforms other models in cognitive diagnosis and matrix factorization in terms of prediction quality and time complexity. There is also an optimal value of the latent factor K and hyperparameter c0 for each dataset. Overall, this paper contributes to the improvement of matrix factorization in educational data.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Hardware & Architecture
Dekui Wang, Jun Feng, Ke Liu, Wei Zhou, Xingxing Hao, Xiaodan Zhang
Summary: This article introduces a fast FPGA connection router called PRoute, which implements a novel prerouting-based parallel local routing algorithm. PRoute precomputes potential routing solutions for various connection patterns on FPGAs, and achieves runtime efficiency and global search ability through parallel local search and A-star maze expansion. Experimental results show that PRoute achieves significant speedups without degrading the quality of results.
IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS
(2023)
Article
Engineering, Multidisciplinary
Mustafa Mhamed, Richard Sutcliffe, Xia Sun, Jun Feng, Ephrem Afele Retta
Summary: Sentiment analysis aims to extract emotions from textual data, and various challenges have emerged due to the proliferation of social media platforms and the flow of data in the Arabic language. This paper introduces Gated Convolution Long (GCL), an architecture designed for Arabic Sentiment Analysis, which overcomes difficulties with lengthy sequence training samples and improves performance for binary and multiple classifications. The proposed method achieves better results than baselines in various Arabic datasets, and includes a Custom Regularization Function (CRF) that enhances performance and optimizes validation loss. Furthermore, the paper explores the relationship between Modern Standard Arabic and five Arabic dialects through a cross-dialect training study.
ENGINEERING SCIENCE AND TECHNOLOGY-AN INTERNATIONAL JOURNAL-JESTECH
(2023)
Article
Computer Science, Artificial Intelligence
Qiang Lu, Xia Sun, Yunfei Long, Zhizezhang Gao, Jun Feng, Tao Sun
Summary: Sentiment analysis (SA) has achieved significant breakthroughs in the past decade and there is a growing interest in multimodal SA (MSA). This article provides a comprehensive overview of SA advances, introduces a novel framework for SA tasks, and discusses the workflows and recent advances of single-modal SA. It also explores the research gaps and challenges in MSA, and proposes potential directions for future improvement.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Chemistry, Multidisciplinary
Ephrem Afele Retta, Richard Sutcliffe, Jabar Mahmood, Michael Abebe Berwo, Eiad Almekhlafi, Sajjad Ahmad Khan, Shehzad Ashraf Chaudhry, Mustafa Mhamed, Jun Feng
Summary: Cross-lingual and multilingual training can be an effective strategy for training an SER classifier when resources for a language are scarce. The difficulty of SER varies for different languages, and better results can be obtained by using two or three non-target languages for training.
APPLIED SCIENCES-BASEL
(2023)
Article
Engineering, Electrical & Electronic
Ke Liu, Dekui Wang, Dongya Wu, Yutao Liu, Jun Feng
Summary: The aim of this research is to improve the performance of human speech emotion recognition. The proposed multi-level attention network (MLAnet) extracts low-level emotion features from the popular mel-scale frequency cepstral coefficient (MFCC) and weights these features using a multi-unit attention module. Experimental results show that this method outperforms other state-of-the-art approaches.
IEEE SIGNAL PROCESSING LETTERS
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Ke Liu, Chen Wang, Jiayue Chen, Jun Feng
Summary: The study proposes a novel Time-Frequency Attention (TFA) method to better extract low-level features in speech emotion recognition and improve accuracy. By utilizing Squeeze-and-Excitation (SE) blocks to effectively integrate global information, the experimental results indicate that the proposed method outperforms existing methods with significant improvements in emotion recognition accuracy.
MULTIMEDIA MODELING (MMM 2022), PT I
(2022)