Article
Computer Science, Interdisciplinary Applications
Yifei Sun, Xin Sun, Zhuo Liu, Yifei Cao, Jie Yang
Summary: This study proposes a novel dynamic community detection algorithm based on particle swarm optimization, targeting the classification of nodes with similar attributes in networks that change over time. By calculating the resistance distance of each node, the core nodes in the network are identified and the constant community is formed by nodes associated with these core nodes. Knowledge gained from the evolution of core nodes in consecutive time steps is utilized to determine the constant community to be retained. Experimental results on various networks indicate the higher accuracy and stability of the proposed algorithm compared to other well-known algorithms.
COMPUTERS & INDUSTRIAL ENGINEERING
(2023)
Article
Chemistry, Physical
Neda Zarayeneh, Nitesh Kumar, Ananth Kalyanaraman, Aurora E. Clark
Summary: This article introduces an algorithm called Delta-screening for identifying temporal communities. The algorithm is flexible in handling the evolving compositions, merging, and splitting behaviors within chemical networks, and is able to resolve multiple time scales.
JOURNAL OF CHEMICAL THEORY AND COMPUTATION
(2022)
Article
Automation & Control Systems
Ismail Koc
Summary: In this study, six different metaheuristic algorithms are utilized to solve community detection problems, with COOT algorithm proving to be the most effective and efficient in terms of solution quality and time.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2022)
Article
Automation & Control Systems
Ismail Koc
Summary: The study utilizes six metaheuristic algorithms to solve community detection problems and proposes a fast method based on CommunityID. Experimental results show that the COOT algorithm is more efficient and faster in solving the problem.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2022)
Article
Multidisciplinary Sciences
Kai Qi, Heng Zhang, Yang Zhou, Yifan Liu, Qingxiang Li
Summary: This study introduces an algorithm called PR-LFM, which combines an improved local fitness maximization (LFM) algorithm with the PageRank (PR) algorithm for community partitioning on cyberspace resources. The experimental data demonstrate good results in the resource division of cyberspace.
SCIENTIFIC REPORTS
(2023)
Article
Engineering, Multidisciplinary
Qiang He, Xin Yan, Xingwei Wang, Tianhang Nan, Zhixue Chen, Xuan He, Min Huang
Summary: This paper studies the Dynamic Opinion Maximization (DOM) problem and proposes an effective hybrid framework to address it. The activated opinion model is designed to estimate the activation status and dynamic opinion process of activated nodes effectively. An effective hybrid framework is proposed for selecting seed nodes, which includes community detection, determination of candidate seed nodes, and a seeding algorithm. Experimental results show the superiority of the proposed framework on average opinions and activated nodes compared to baseline algorithms.
IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING
(2023)
Article
Computer Science, Information Systems
Ying Yin, Yuhai Zhao, He Li, Xiangjun Dong
Summary: The paper proposes an efficient and effective multi-objective method, DYN-MODPSO, which addresses the issues in dynamic community detection by enhancing the traditional evolutionary clustering framework and particle swarm algorithm. The novel strategy and carefully designed operators contribute to the method's superior performance on both real and synthetic dynamic networks, outperforming competitors in terms of effectiveness and efficiency.
INFORMATION SCIENCES
(2021)
Article
Computer Science, Artificial Intelligence
Sajjad Molaei, Hadi Moazen, Samad Najjar-Ghabel, Leili Farzinvash
Summary: The article introduces a new variant of PSO algorithm, PSOLC, which enhances its search performance through improved learning strategy and crossover operator. By altering exemplar particles, updating parameters, and integrating with genetic algorithm, the algorithm shows significant improvements in exploration and exploitation capabilities.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Salihu A. Abdulkarim, Andries P. Engelbrecht
Summary: Several studies have applied particle swarm optimization algorithms to train neural networks for time series forecasting, with good performance results. This study introduces a dynamic PSO algorithm for training NNs in forecasting non-stationary time series, outperforming standard PSO and Rprop algorithms. These findings suggest the potential of dynamic PSO in real-world forecasting applications.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Multidisciplinary Sciences
Alexandre Bovet, Jean-Charles Delvenne, Renaud Lambiotte
Summary: This article introduces a method based on a dynamical process evolving on a temporal network, which uncovers different dynamic scales in a system by considering the ordering of edges in forward and backward time. The method provides a new approach to extracting a simplified view of time-dependent network interactions in a system.
Article
Mathematical & Computational Biology
Changwei Gong, Bing Xue, Changhong Jing, Chun-Hui He, Guo-Cheng Wu, Baiying Lei, Shuqiang Wang
Summary: This paper proposes a time-sequential graph adversarial learning (TGAL) framework for detecting brain communities and characterizing the structure of communities in brain networks. The framework utilizes a novel time-sequential graph neural network as an encoder to extract efficient graph representations using spatio-temporal attention mechanism. The effectiveness of the framework is demonstrated through experiments on real-world brain network datasets, showcasing its advantage in brain community detection.
MATHEMATICAL BIOSCIENCES AND ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Xu Yang, Hongru Li, Xia Yu
Summary: This paper proposes a dynamic multi-swarm cooperation PSO with dimension mutation (MSCPSO) to overcome the shortcomings of PSO in solving complex optimization problems. The adaptive sample selection strategy (ASS) and adaptive dimension mutation strategy (ADM) are the two contributions of MSCPSO. Experimental results show that MSCPSO outperforms other methods in most complex and multimodal conditions.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2022)
Article
Computer Science, Artificial Intelligence
Qinghua Gu, Qian Wang, Lu Chen, Xiaoguang Li, Xuexian Li
Summary: This paper proposes a new multi-objective particle swarm optimization algorithm with a dynamic neighborhood balancing mechanism (DNB-MOPSO) to solve the multi-modal multi-objective optimization problems with the same fitness value for Pareto-optimal solutions. The algorithm balances local and global search using an adaptive parameter adjustment strategy and employs a mutation operator to escape from local optima. Additionally, a dynamic neighborhood reform strategy based on current niching methods is implemented to enhance exploration and maintain population diversity. Experimental results demonstrate the superiority of the proposed algorithm in locating more optimal solutions in the decision space.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Hadi Moazen, Sajjad Molaei, Leili Farzinvash, Masoud Sabaei
Summary: This paper proposes a particle swarm optimization algorithm called PSO-ELPM, which balances the exploration and exploitation capabilities of PSO through elite learning, enhanced parameter updating, and exponential mutation operator. The algorithm uses the best-performing particles as exemplars to guide the optimization process and computes self-cognition coefficients based on these elites. It also ensures a smooth distribution of weight among the elites using the inverse of the cube root function, and applies an exponential mutation operator to determine the mutation probability per particle. Comparisons with 10 state-of-the-art PSO variants on the CEC 2017 benchmark functions show that the proposed algorithm achieves higher accuracy with acceptable time complexity.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Linqiang Pan, Yi Zhao, Lianghao Li
Summary: The study introduces a new neighborhood-based particle swarm optimization algorithm MNPSO, which can locate multiple roots of nonlinear equation systems in a single run. Experimental results demonstrate that MNPSO outperforms other methods in terms of root ratio and success ratio.
SWARM AND EVOLUTIONARY COMPUTATION
(2022)
Article
Computer Science, Information Systems
Lei Xiao, Junhong Feng, Xishuan Niu, Jian-Hong Wang
Summary: Developing sensor ontologies and using matching techniques to annotate sensor data is a feasible approach to addressing data heterogeneity in IoT. This study proposes a competitive binary particle swarm optimization algorithm for sensor ontology matching and demonstrates its superiority over other swarm intelligence methods through experiments.
MOBILE INFORMATION SYSTEMS
(2022)
Article
Biochemical Research Methods
Ju Xiang, Xiangmao Meng, Yichao Zhao, Fang-Xiang Wu, Min Li
Summary: In this study, a hybrid disease-gene prediction method called HyMM is proposed, which utilizes multiscale module structure to enhance the prediction of disease-related genes. HyMM extracts module partitions at different scales and estimates gene-disease relatedness based on the abundance of disease-related genes within the modules. The results confirm the stable and good performance of HyMM compared to other state-of-the-art methods and demonstrate the further performance improvement achieved through parameter estimation.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Computer Science, Artificial Intelligence
Rayyan Azam Khan, Yigang Luo, Fang-Xiang Wu
Summary: This study proposes a deep learning-based multi-scaled generative adversarial network (GAN) for processing images with heterogeneous blur. By concatenating images of different scales and using residual image learning, the method is effective in reducing blur while preserving structural properties. Experimental results demonstrate its superior performance in image analysis.
IET IMAGE PROCESSING
(2022)
Article
Biochemical Research Methods
Zhongjian Cheng, Cheng Yan, Fang-Xiang Wu, Jianxin Wang
Summary: This study proposes an end-to-end deep learning method called MHSADTI for predicting drug-target interactions (DTIs). The method uses a graph attention network and multi-head self-attention mechanism to extract features from drugs and proteins. Attention scores are then used to determine important amino acid subsequences in proteins for predicting drug interactions. Experimental results show that MHSADTI outperforms state-of-the-art methods in multiple metrics and provides effective visualization for interpreting prediction results.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2022)
Article
Biochemical Research Methods
Xingyi Li, Ju Xiang, Fang-Xiang Wu, Min Li
Summary: This study developed a multiplex network-based dual ranking framework for analyzing heterogeneous complex diseases. The results showed that the proposed method could identify biomarkers with small quantity, great prediction performance, and biological interpretability, and outperformed other competing methods in terms of diagnosis, prognosis, and classification.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2022)
Article
Multidisciplinary Sciences
Ali Akbar Jamali, Yuting Tan, Anthony Kusalik, Fang-Xiang Wu
Summary: Computational drug repositioning using tensor decomposition can accelerate drug discovery process. NTD-DR, a nonnegative tensor decomposition method, outperforms other methods in prediction performance and is validated in case studies.
Article
Engineering, Multidisciplinary
Xiangmao Meng, Wenkai Li, Ju Xiang, Hayat Dino Bedru, Wenkang Wang, Fang-Xiang Wu, Min Li
Summary: This study reexamines the essentiality of hub proteins in PPI networks by constructing temporal-spatial dynamic PPI networks and integrating gene expression data and subcellular localization information. The results show that integrating multiple data sources can improve the identification accuracy of essential proteins.
IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING
(2022)
Article
Biochemical Research Methods
Yulian Ding, Xiujuan Lei, Bo Liao, Fang-Xiang Wu
Summary: In this study, a factorization machine-based deep neural network with binary pairwise encoding (DFMbpe) is proposed to identify disease-related biomarkers. The DFMbpe model considers the interdependence of features and combines low-order and high-order feature interactions, leading to better performance compared to other biomarker identification models.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2023)
Article
Computer Science, Hardware & Architecture
Junhong Feng, Jie Zhang, Xiaoshu Zhu, Jian-Hong Wang
Summary: The study proposes a novel gene selection algorithm based on Fisher score and genetic algorithms with dynamic crossover (FDCGA) for data clustering in single-cell RNA sequencing (scRNA-seq) data, demonstrating its superior performance compared to other competing methods.
JOURNAL OF SUPERCOMPUTING
(2023)
Article
Engineering, Biomedical
Rayyan Azam Khan, Yigang Luo, Fang-Xiang Wu
Summary: In this study, a multilevel generative adversarial network (GAN) is proposed to enhance computed tomographic (CT) images for liver cancer diagnosis. The performance of the proposed method is investigated using three publicly available datasets, and it achieves good results in terms of performance metrics and computer-aided diagnosis. The effectiveness of the proposed multi-level GAN in producing enhanced biomedical images with preserved structural details and reduction in artifacts is demonstrated, and it shows consistently better performance among three datasets for computer-aided diagnosis.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2023)
Article
Computer Science, Information Systems
Minghan Fu, Meiyun Wang, Yaping Wu, Na Zhang, Yongfeng Yang, Haining Wang, Yun Zhou, Yue Shang, Fang-Xiang Wu, Hairong Zheng, Dong Liang, Zhanli Hu
Summary: A novel two-branch network architecture called SW-GCN is proposed to improve PET image quality. The network utilizes Swin Transformer units and graph convolution operation to handle different types of input information flow and enables better processing of long-range contextual information. Experimental results demonstrate that the proposed approach outperforms state-of-the-art methods in both quantitative and qualitative evaluations.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2023)
Article
Chemistry, Analytical
Kawsar Ahmed, Francis M. Bui, Fang-Xiang Wu
Summary: To reduce the development time and effort of standard optical biosensors, machine learning approaches have been used to predict crucial parameters and evaluate the performance of the models based on performance indicators.
Article
Computer Science, Information Systems
Yuchen Zhang, Xiujuan Lei, Cai Dai, Yi Pan, Fang-Xiang Wu
Summary: More and more studies have shown that circRNAs can be used as disease markers due to their stability. Various computational methods, particularly those utilizing artificial intelligence, have been employed to predict circRNA-disease associations. However, these methods often use single, standard objective functions, leading to low prediction accuracy. This paper proposes a multiobjective evolutionary algorithm called ICDMOE to identify circRNA-disease associations, using matrix factorization and modularity of similarity networks to design four objective functions. Experimental results demonstrate that ICDMOE outperforms other prediction methods and can provide good candidates for biomedical experiments, as confirmed by existing studies, miRNA regulations, and expression profiles.
INFORMATION SCIENCES
(2023)
Article
Biochemical Research Methods
Xuhua Yan, Ruiqing Zheng, Fangxiang Wu, Min Li
Summary: CIAIRE is a novel contrastive learning-based batch correction framework that achieves a superior mix-heterogeneity trade-off. It proposes two complementary strategies, construction strategy and refinement strategy, to improve the appropriateness of positive pairs. CLAIRE outperforms existing methods in terms of mix-heterogeneity trade-off and achieves the best integration performance on six real datasets.
Article
Biochemical Research Methods
Yiming Li, Min Zeng, Fuhao Zhang, Fang-Xiang Wu, Min Li
Summary: In this study, DeepCellEss, a sequence-based interpretable deep learning framework, is proposed for cell line-specific essential protein predictions. By utilizing convolutional neural networks, bidirectional long short-term memory, and multi-head self-attention mechanism, DeepCellEss achieves effective prediction performance for different cell lines and outperforms existing methods and metrics.
