Article
Environmental Sciences
Jingjing Cai, Fengming Gan, Xianghai Cao, Wei Liu, Peng Li
Summary: This study proposes a self-supervised learning framework called CL-CNN for radar signal intra-pulse modulation classification. By using a two-stage training strategy, the model achieves excellent performance on classification tasks and exhibits good generalization ability.
Article
Engineering, Electrical & Electronic
Yifan Shen, Ling Shi, Ji Zhao, Yuting Dong, Lizhe Wang
Summary: This paper proposes a fully convolutional spectral-spatial fusion network based on supervised contrastive learning for hyperspectral image classification, aiming to enhance the classification performance by fusing spectral and spatial information.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Telecommunications
Xue Fu, Guan Gui, Yu Wang, Tomoaki Ohtsuki, Bamidele Adebisi, Haris Gacanin, Fumiyuki Adachi
Summary: This paper proposes a decentralized learning AMC method using model consolidation and lightweight design, which reduces the storage and computational capacity requirements, improves the training efficiency, and lowers the communication overhead.
IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING
(2022)
Article
Telecommunications
Hao Zhang, Fuhui Zhou, Qihui Wu, Wei Wu, Rose Qingyang Hu
Summary: A novel automatic modulation classification scheme inspired by face recognition and utilizing a multi-scale network was proposed, with a new loss function combining center loss and cross entropy loss to learn discriminative and separable features for improved classification performance. Extensive simulation results showed that the proposed scheme outperformed benchmark schemes in terms of classification accuracy.
IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING
(2022)
Article
Engineering, Electrical & Electronic
Wensheng Lin, Dongbin Hou, Junsheng Huang, Lixin Li, Zhu Han
Summary: This paper proposes a transfer learning model for automatic modulation recognition using only a few samples. By training the model with a combination of audio and modulated signals, and fine-tuning it with limited samples, the classification accuracy is significantly improved.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Indrajit Mazumdar, Jayanta Mukherjee
Summary: Automatically segmenting brain tumors from MRI scans is challenging due to their diverse characteristics. This study presents an accurate and efficient CNN model, called ESA-Net, for fully automatic brain tumor segmentation. The proposed model outperforms traditional methods in segmentation accuracy while achieving faster inference speed and lower parameter count.
Article
Telecommunications
Weisi Kong, Xun Jiao, Yuhua Xu, Bolin Zhang, Qinghai Yang
Summary: The application of deep learning improves the speed and accuracy of automatic modulation recognition (AMR), enabling intelligent spectrum management and electronic reconnaissance. However, deep learning-aided AMR often requires a large number of labelled samples, which are limited in practical applications. This paper proposes a Transformer-based contrastive semi-supervised learning framework for AMR, which uses unlabeled samples for pre-training and labelled samples for fine-tuning. It also introduces a convolutional transformer deep neural network to address the challenges in applying Transformer to AMR. Experimental results demonstrate the feasibility, superiority, and stability of the proposed framework.
IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING
(2023)
Article
Telecommunications
Jiawei Zhang, Tiantian Wang, Zhixi Feng, Shuyuan Yang
Summary: With the advancement of modern communications technology, automatic modulation classification (AMC) has become increasingly important in complex wireless communication environments. Existing deep learning-based AMC schemes have limitations in utilizing feature maps, whereas the proposed adaptive wavelet network (AWN) overcomes this limitation by introducing adaptive wavelet decomposition and channel attention mechanism. AWN explicitly extracts features from multiple frequency bands and selects optimal frequencies, efficiently integrating signal properties in the frequency domain. Simulation results demonstrate the superiority of the proposed AMC scheme over benchmark schemes in terms of performance and computational complexity.
IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING
(2023)
Article
Computer Science, Information Systems
Shuo Chang, Sai Huang, Ruiyun Zhang, Zhiyong Feng, Liang Liu
Summary: Automatic modulation classification (AMC) plays a vital role in identifying the modulation type of a received signal for ensuring the physical-layer security of IoT networks. This article focuses on reproducing and evaluating popular AMC algorithms using the in-phase/quadrature (I/Q) and amplitude/phase (A/P) representations for comparison. Based on the experimental results, it is found that CNN-RNN-like algorithms using A/P as input data perform better at high signal-to-noise ratio (SNR), while the opposite is true at low SNR. Inspired by these findings, a multitask learning-based deep neural network (MLDNN) is proposed, which effectively fuses I/Q and A/P. Extensive simulations demonstrate the superior performance of the proposed MLDNN in a public benchmark.
IEEE INTERNET OF THINGS JOURNAL
(2022)
Article
Engineering, Electrical & Electronic
Chenghong Xiao, Shuyuan Yang, Zhixi Feng
Summary: In this article, a novel end-to-end automatic modulation classification (AMC) model called complex-valued depthwise separable convolutional neural network (CDSCNN) is proposed, which adopts complex-valued operation units for tailored feature learning for AMC. With an overall accuracy of 62.63% on the RadioML2016.10a dataset, CDSCNN outperforms its counterparts by 1%-11%. After fine-tuning on the RadioML2016.10b dataset, the overall accuracy reaches 63.15%, demonstrating the robust recognition and generalization capability of CDSCNN. Moreover, CDSCNN exhibits lower model complexity compared to other methods.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Chemistry, Analytical
Dong Wang, Meiyan Lin, Xiaoxu Zhang, Yonghui Huang, Yan Zhu
Summary: In this study, a CNN-transformer graph neural network (CTGNet) is proposed for modulation classification, which aims to uncover complex representations in signal data by transforming them into graph structures. Extensive experiments demonstrate that our method outperforms advanced deep learning techniques and achieves the highest recognition accuracy, highlighting the significant advantage of CTGNet in capturing key features in signal data and providing an effective solution for modulation classification tasks.
Article
Environmental Sciences
Mengyu Yang, Wensi Wang, Qiang Gao, Chen Zhao, Caole Li, Xiangfei Yang, Jiaxi Li, Xiaoguang Li, Jianglong Cui, Liting Zhang, Yanping Ji, Shuqin Geng
Summary: This research utilizes deep learning techniques and transfer learning to automatically classify and identify harmful phytoplankton. The results demonstrate that transfer learning can significantly improve the recognition performance of harmful phytoplankton, and the proposed method is effective in preliminary screening and reduces the workload of professionals.
ENVIRONMENTAL SCIENCE AND POLLUTION RESEARCH
(2023)
Article
Environmental Sciences
Qian Liu, Zebin Wu, Xiuping Jia, Yang Xu, Zhihui Wei
Summary: A class feature fused fully convolutional network (CFF-FCN) is proposed, which incorporates a local feature extraction block and a class feature fusion block to utilize local and global information. Experimental results demonstrate the superiority of the network over other deep learning methods, especially in cases with a small number of training samples.
Article
Computer Science, Information Systems
Han Yang, Jun Li
Summary: In this paper, a prototypical contrastive learning (ProCL) is proposed for image classification by combining contrastive learning and clustering. ProCL performs representation learning by clustering semantically similar images into the same group and encouraging clustering consistency between different augmentations of the same image. Negative samples are weighted according to the distance between the prototypes, resulting in more effective performance.
CLUSTER COMPUTING-THE JOURNAL OF NETWORKS SOFTWARE TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Hany S. Hussein, Mohamed Hassan Essai Ali, Mohammed Ismeil, Mohamed N. Shaaban, Mona Lotfy Mohamed, Hany A. Atallah
Summary: This study proposes robust CNN-based AMC techniques, which eliminate the need for feature extraction and can achieve high classification accuracy. The developed techniques utilize different classification layers and can automatically learn the features from the transmitted signals during training.
Article
Computer Science, Hardware & Architecture
Rafiul Haq, Xiaowang Zhang, Wahab Khan, Zhiyong Feng
Summary: Named Entity Recognition (NER) is fundamental for various natural language processing tasks, and while English NER systems have advanced, Urdu NER systems are still in their early stages. This study proposes deep neural approaches that automatically learn features and eliminate manual feature engineering, resulting in notable progress in Urdu NER systems.
Article
Engineering, Electrical & Electronic
Zhishu Qu, James R. Kelly, Zhengpeng Wang, Shaker Alkaraki, Yue Gao
Summary: This letter presents a novel reconfigurable microstrip patch antenna that switches its main beam direction using liquid metal. The antenna utilizes both the parasitic steering approach and a novel switchable ground plane. The antenna operates at 5.9 GHz and consists of a driven patch surrounded by four parasitics. The ground plane incorporates two reconfigurable segments formed from liquid metal.
IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS
(2023)
Article
Computer Science, Software Engineering
Xinyue Zhou, Zhiyong Feng, Jianmao Xiao, Shizhan Chen, Xiao Xue, Hongyue Wu
Summary: With the comprehensive interconnection in human-cyber-physical systems, e-services are growing rapidly in platform retailing. E-services as the main body of transactions have begun to dominate the construction of retail platforms. This article introduces the concept of Service as a Commodity (SaaC) and discusses its value factors, principles, and proposed framework for service integration and management. A real case of aging healthcare is also presented to demonstrate its practical value. SaaC leads the evolution of the service ecosystem and comprehensively enhances market activity.
IEEE INTERNET COMPUTING
(2023)
Article
Engineering, Electrical & Electronic
Yue Cao, Shaoshi Yang, Zhiyong Feng, Lihua Wang, Lajos Hanzo
Summary: A distributed spatio-temporal information based cooperative positioning (STICP) algorithm is proposed for wireless networks operating in GNSS denied environments. The algorithm supports any ranging measurements that can determine the distance between nodes. It utilizes a scaled unscented transform method for approximating nonlinear terms and an enhanced anchor upgrading mechanism to improve computational efficiency.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
(2023)
Article
Telecommunications
Zhiqing Wei, Hanyang Qu, Wangjun Jiang, Kaifeng Han, Huici Wu, Zhiyong Feng
Summary: Integrated sensing and communication (ISAC) is a key technology in the fifth generation advanced (5G-A) and sixth generation (6G) mobile communication systems, and it has the advantages of high spectrum efficiency and low hardware cost. This paper proposes a phase coding method for ISAC signal design to improve its anti-noise performance, and introduces iterative ISAC signal processing methods with low computational complexity to enhance sensing accuracy and energy efficiency.
IEEE TRANSACTIONS ON GREEN COMMUNICATIONS AND NETWORKING
(2023)
Article
Computer Science, Software Engineering
Sofonias Yitagesu, Zhenchang Xing, Xiaowang Zhang, Zhiyong Feng, Xiaohong Li, Linyi Han
Summary: This article proposes an unsupervised method to label and extract important vulnerability concepts in textual vulnerability descriptions (TVDs). The method is based on the observation that phrases of the same type usually share syntactically similar paths in the sentence parsing trees. The article also introduces a concept extraction model to demonstrate the utility of the unsupervisedly labeled concepts.
ACM TRANSACTIONS ON SOFTWARE ENGINEERING AND METHODOLOGY
(2023)
Article
Telecommunications
Shanchuan Ying, Sai Huang, Shuo Chang, Zheng Yang, Zhiyong Feng, Ningyan Guo
Summary: In this paper, a data-driven framework called CTDNN is proposed to enhance the classification performance of automatic modulation classification. By using modules such as convolutional neural network, transition module, and final classifier, CTDNN achieves superior classification performance compared to traditional deep models.
CHINA COMMUNICATIONS
(2023)
Article
Computer Science, Hardware & Architecture
Ping Zhang, Heng Yang, Zhiyong Feng, Yanpeng Cui, Jincheng Dai, Xiaoqi Qin, Jinglin Li, Qixun Zhang
Summary: In this article, a purposeful machine communication framework enabled by JCSC technology is proposed to achieve the vision of intelligent connection of everything towards 6G. The paradigm of wireless communication design is shifted from naive maximalist approaches to intelligent value-based approaches, by fully exploiting the collective intelligence of networked machines. However, there are technical barriers that need to be addressed before the widespread adoption of purposeful communications, including the conception of machine purpose, fast and concise networking strategy, and semantics-aware information exchange mechanism during task-oriented cooperation. Hence, enabling technologies and open challenges are discussed in this paper. Simulation results demonstrate that the proposed framework can significantly reduce networking overhead and improve communication efficiency.
IEEE WIRELESS COMMUNICATIONS
(2023)
Article
Computer Science, Information Systems
Yanwei Xu, Zhiyong Feng, Xian Zhou, Meng Xing, Hongyue Wu, Xiao Xue, Shizhan Chen, Chao Wang, Lianyong Qi
Summary: In this paper, the authors propose an attention-based neural network model called GainTrust for trust relationship prediction in online social networks. They consider the complementary user data of trusted neighbors and temporal continuity of user behaviors, and employ a heterogeneous network embedding layer and a multi-layer LSTM network to learn the features. They also design a two-level multi-head attention mechanism to obtain global trust features. Extensive experiments on real-world datasets validate the effectiveness of the proposed approach.
INFORMATION SCIENCES
(2023)
Article
Chemistry, Analytical
Shanchuan Ying, Sai Huang, Shuo Chang, Jiashuo He, Zhiyong Feng
Summary: In this paper, a dual-task neural network called AMSCN is proposed, which can classify both the modulation and transmitter of the received signal simultaneously. Experimental results show that the AMSCN achieves good performance gains for the SEI task.
Article
Computer Science, Hardware & Architecture
Kan Yu, Jiguo Yu, Zhiyong Feng, Min Guo
Summary: Physical layer security is an effective solution for secure and confidential information transmission in wireless networks, and secrecy transmission capacity (STC) is crucial for analyzing its impact on system parameters. This paper proposes two secrecy improvement strategies, connection guard zone (CGZ) and improved interferer protected zone (IIPZ), which do not consider the channel state information (CSI) of eavesdroppers. The analytical framework of STC is established using stochastic geometry, and conditions for achieving positive STC are derived. The impact of Random WayPoint (RWP) mobile receiver and active eavesdroppers on physical layer security is also considered, and simulations show the performance superiority of IIPZ in reliability and CGZ in secrecy, while RWP mobile destination achieves higher reliability and STC compared to static scenarios.
Article
Computer Science, Software Engineering
Guodong Fan, Shizhan Chen, Hongyue Wu, Cuiyun Gao, Jianmao Xiao, Xiao Xue, Zhiyong Feng
Summary: Software collaborative platforms are crucial for software maintenance, and automatic dialogue summarization is a useful tool for extracting and sharing knowledge. However, the lack of labeled data for noisy dialogues leads to poor performance in few-shot scenarios. To address this, we propose ADSum, a novel approach based on pre-trained models. We fine-tune the T5 model using discussion posts from GitHub and employ the prompt tuning paradigm to improve performance. Experimental results show that our approach achieves state-of-the-art performance and outperforms other models on the GitHub dataset.
JOURNAL OF SYSTEMS AND SOFTWARE
(2023)
Article
Computer Science, Information Systems
Xingjian Zhang, Yuan Ma, Yaohui Liu, Shaohua Wu, Jian Jiao, Yue Gao, Qinyu Zhang
Summary: The study proposes a deep neural network-based robust alternating direction method of multipliers (R-ADMM) to efficiently recover wideband spectrum signals with low signal-to-noise ratio (SNR). By optimizing the learnable parameters and operations of signal reconstruction and adopting numerical differential-based gradient computation, the proposed method achieves significantly improved noise robustness in real-world and simulated signals.
IEEE WIRELESS COMMUNICATIONS LETTERS
(2023)
Article
Computer Science, Information Systems
Xiao Xue, Deyu Zhou, Fangyi Chen, Xiangning Yu, Zhiyong Feng, Yucong Duan, Lin Meng, Mu Zhang
Summary: With the advancement of ICT and service economy, service ecosystem has emerged in various fields. The evolution of service ecosystem is influenced by the interaction of three heterogeneous networks - social network, service network, and value network. Traditional analysis models based on Service Oriented Architecture (SOA) are inadequate in this scenario, making the analysis of service ecosystem a challenging task. This paper introduces a value oriented analysis framework (VOA) that utilizes value as a key factor in describing the interaction between the three networks. Additionally, a computational experiment system is developed to test the effectiveness of the VOA framework and explore the impact of different intervention strategies. The results demonstrate that the proposed analysis framework provides new methods and ideas for analyzing service ecosystems.
IEEE TRANSACTIONS ON SERVICES COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Dongxiao He, Tao Wang, Lu Zhai, Di Jin, Liang Yang, Yuxiao Huang, Zhiyong Feng, Philip S. Yu
Summary: Network embedding, a technique for learning low dimensional node representations in networks, has been widely used in network analysis. However, existing methods based on Generative Adversarial Networks (GAN) struggle to distinguish node representations from Gaussian distribution. To address this issue, we propose a novel adversarial learning framework called ArmGAN, which applies adversarial learning strategy on the representation mechanism. Experimental results show that ArmGAN outperforms existing methods on various tasks.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
