2023 Vol. 32, No. 4

REVIEW
Characteristic Mode Analysis: Application to Electromagnetic Radiation, Scattering, and Coupling Problems
DENG Xuan, ZHANG Di, CHEN Yikai, YANG Shiwen
2023, 32(4): 663-677. doi: 10.23919/cje.2022.00.200
Abstract(1149) HTML (565) PDF(250)
Abstract:
Recent years, the theory of characteristic modes has emerged as a powerful analysis technique in antenna engineering, providing a means to reveal the natural resonant properties of objects and providing a variety of modal parameters. Based on these modal parameters, advanced techniques have been developed based on the theory of characteristic modes to address a wide range of electromagnetic radiation, scattering, and coupling problems. This review provides an overview of some of the latest characteristic mode-based techniques for wideband design, circular polarization, radiation pattern control, scattering control, and mutual coupling control. In addition, future perspectives are discussed, highlighting the potential of characteristic modes for addressing even more complex electromagnetics problems.
ELECTROMAGNETICS AND MICROWAVE
Coupling Enhancement of THz Metamaterials Source with Parallel Multiple Beams
ZHANG Kaichun, FENG Yuming, ZHAO Xiaoyan, HU Jincheng, XIONG Neng, GUO Sidou, TANG Lin, LIU Diwei
2023, 32(4): 678-682. doi: 10.23919/cje.2022.00.032
Abstract(406) HTML (206) PDF(59)
Abstract:
In this paper, we propose a terahertz radiation source over the R-band (220–325 GHz) based on metamaterials (MTMs) structure and parallel multiple beams. The effective permittivity and permeability of the slow-wave structure can be obtained through the S-parameter retrieval approach, using numerical simulation. Additionally, the electromagnetic properties of the MTMs structure are analyzed, including the dispersion and the coupling impedance. Furthermore, we simulate the beam-wave interaction of the backward oscillator (BWO) with MTMs structure and parallel multiple beams by 3-D particle-in-cell code. It is observed that parallel multiple beams can highly enhance the beam-wave interaction and greatly enlarge the output power. These results indicate that the saturated (peak) output power is approximately 63W with the efficiency of roughly 6% at the operating frequency of 231 GHz, under the beam voltage of 35 kV and total current of 30 mA (6-beam) respectively. Meanwhile, the BWO can generate power of 10–80 W in the tunable frequency of 220–240 GHz.
A Novel Wideband Wilkinson Pulse Combiner with Enhanced Low Frequency Isolation
WANG Zitong, WU Qi, SU Donglin
2023, 32(4): 683-691. doi: 10.23919/cje.2021.00.429
Abstract(620) HTML (308) PDF(71)
Abstract:
In this paper, a novel Wilkinson pulse combiner (WPC) is proposed for the combination of Gaussian pulse signals. The WPC requires a very wide bandwidth, small size and high port isolation. To improve the operating bandwidth, the design adopts the form of eight-section WPC. Eight capacitors are connected in series with the isolating resistors of each section. After capacitive loading, isolation between WPC input ports is significantly improved at low frequency. Consequently, the operating bandwidth of WPC has been increased from 13:1 to 31:1. Compared with the conventional Wilkinson combiner with the same bandwidth, the proposed WPC reduces the size by 40%. In addition, all the ports are well impedance matched and the insertion loss in the operating frequency band is less than 0.5 dB. To verify the feasibility of the design, a prototype was fabricated and measured. Experiment shows that the novel WPC is more advantageous to generate dual-Gaussian pulse signals.
Dual-Band Flexible MIMO Antenna with Self-Isolation Enhancement Structure for Wearable Applications
YANG Lingsheng, XIE Yizhang, JIA Hongting, QU Meixuan, LU Zhengyan, LI Yajie
2023, 32(4): 692-702. doi: 10.23919/cje.2021.00.293
Abstract(347) HTML (176) PDF(46)
Abstract:
A two-element dual-band flexible multi-in multi-out antenna which can be used for wearable applications is proposed in this paper. The antenna consists of two radiating elements fed by coplanar waveguide, and a shielding layer, which are all made of flexible conductive cloth MKKTN260. Each radiating element is composed of two coupled split ring-shaped bending strips. The proposed antenna shows two measured impedance bandwidth (S11<−10 dB) of 2.39–2.48 GHz and 5.72–5.88 GHz, so that it can be used for 2.4 GHz and 5.8 GHz ISM (industrial scientific medical) applications. The two coupled split rings form a self-isolation enhancement structure and can realize polarization diversity at 2.4 GHz band and radiation shielding at 5.8 GHz band, respectively. High isolation (>30 dB) has been achieved for both the bands. Other characteristics for wearable applications like gain, efficiency, specific absorption rate, and bending performances were also studied.
Design of Pyramidal Horn with Arbitrary E/H Plane Half-Power Beamwidth
ZHANG Wenrui, SHAO Wenyuan, JI Yicai, LI Chao, YANG Guan, LU Wei, FANG Guangyou
2023, 32(4): 703-709. doi: 10.23919/cje.2021.00.212
Abstract(905) HTML (423) PDF(52)
Abstract:
This paper proposes a novel design method for pyramid horns which are under the constraints of 3 dB beamwidth. It is based on the general radiation patterns of E/H planes derived from Huygens’ principle. Through interpolation and fitting techniques, the E/H plane’s maximum aperture error parameter of the pyramid horn is obtained as a function of the angle and aperture electrical size. Firstly, the aperture size of the E (or H) plane is calculated with the help of the optimal gain principle. Secondly, the constraint equation of another plane is derived. Finally, the intersection of constraint equation and interpolation function, which can be solved iteratively, contains all the solution information. The general radiation patterns neglect the influence of the Huygens element factor which makes the error bigger in large design beamwidth. In this paper, through theoretical analysis and simulation experiments, two correction formulas are employed to correct the Huygens element factor’s influence on the E/H planes. Simulation experiments and measurements show that the proposed method has a smaller design error in the range of 0–60 degrees half-power beamwidth.
Dual Radial-Resonant Wide Beamwidth Circular Sector Microstrip Patch Antennas
MAO Xiaohui, LU Wenjun, JI Feiyan, XING Xiuqiong, ZHU Lei
2023, 32(4): 710-719. doi: 10.23919/cje.2021.00.219
Abstract(787) HTML (384) PDF(101)
Abstract:
In this article, a design approach to a radial-resonant wide beamwidth circular sector patch antenna is advanced. As properly evolved from a U-shaped dipole, a prototype magnetic dipole can be fit in the radial direction of a circular sector patch radiator, with its length set as the positive odd-integer multiples of one-quarter wavelength. In this way, multiple ${\boldsymbol{{\bf{TM}}_{0m}(m=1,}}$${\boldsymbol{2,...)}}$ modes resonant circular sector patch antenna with short-circuited circumference and widened E-plane beamwidth can be realized by proper excitation and perturbations. Prototype antennas are then designed and fabricated to validate the design approach. Experimental results reveal that the E-plane beamwidth of a dual-resonant antenna fabricated on air/Teflon substrate can be effectively broadened to 128°/120°, with an impedance bandwidth of 17.4%/7.1%, respectively. In both cases, the antenna heights are strictly limited to no more than 0.03-guided wavelength. It is evidently validated that the proposed approach can effectively enhance the operational bandwidth and beamwidth of a microstrip patch antenna while maintaining its inherent low profile merit.
Multiple Plasmonic Fano Resonances Revisited with Modified Transformation-Optics Theory
JIANG Jing, LIANG Mingli, LI Jiaying
2023, 32(4): 720-730. doi: 10.23919/cje.2022.00.095
Abstract(359) HTML (184) PDF(53)
Abstract:
Plasmonic Fano resonances have recently attracted a great deal of research interest due to their sharp asymmetric profile, high sensitivity to the ambient material and low radiation damping. In this paper, we extend the plasmonic Fano resonances that resulted from the interference between bright and dark modes in a core-shell nanoparticle system for novel biosensing, optical switching applications, by using the theory of transformation optics. To this end, we consider the optical properties of dielectric-core-metallic-shell dimer and derive full analytical formulae for different geometric configurations. Our results demonstrated that breaking the geometrical symmetry of the structure, multiple Fano resonances arise owing to the near-field coupling of the bright and dark resonant modes. Electromagnetic induced transparency (EIT)-like effects are observed when the resonance frequencies of the bright and dark modes overlap. Strong dependence of the localized surface plasmon (LSP) on the geometry renders the multiple Fano resonances highly tunable in the proposed structure through independently altering the spectral profile of each nanoparticle, which provides much feasibility since most multiple Fano resonances reported in literature are results of collective plasmonic behavior and cannot be tuned independently. Furthermore, the figure of merit for refractive index sensing of the higher-order dark modes are predicted to be up to 36% greater than that of a single nanowire. These results make the proposed nanostructure attractive for many potential applications such as multiwavelength biosensing, switching and modulation.
Design and Realization of Broadband Active Inductor Based Band Pass Filter
Aysu Belen, Mehmet A. Belen, Merih Palandöken, Peyman Mahouti, Özlem Tari
2023, 32(4): 731-735. doi: 10.23919/cje.2021.00.322
Abstract(524) HTML (258) PDF(59)
Abstract:
With the latest developments in the wireless communication systems, the alternative design methodologies are required for the broadband design of microwave components. In this paper, a compact broad band pass filter (BPF) design is introduced through the microwave design technique based on the active inductor (AIN) with the numerical computation and experimental measurement studies. The proposed AIN based BPF has operating frequency band extending from 0.8 GHz to 2.7 GHz in compact size with high selectivity in comparison to conventional LC based BPF. The experimental measurement results agree well with the numerical computation results. The proposed AIN based BPF design has technical capability to be conveniently tuned to operate at different frequency bands.
Optimized Design of Multi-Layer Absorber for Human Tissue Surface
TU Botao, YE Mengqiu, YANG Zhen, LI Jinfeng, LI Guanghui, ZHANG Yuejin
2023, 32(4): 736-746. doi: 10.23919/cje.2021.00.385
Abstract(456) HTML (225) PDF(55)
Abstract:
In this paper, the structure optimization scheme of multi-layer absorber on the surface of human tissue is designed. The absorber uses graphite, foam and other materials to build a resistance loss layer. Solve the electromagnetic parameters of graphite through its characteristics, use the equivalent transmission line theory to calculate the reflection coefficient. Establish the objective function of the reflection coefficient, and use genetic algorithm to optimize the design of the absorbing device. The experimental results show that compared with the Jaumann type three-layer absorber, the reflection coefficient of the multi-layer absorber optimized by genetic algorithm in this paper has decreased by nearly 13 dB. From the analysis of error and sensitivity, it can be concluded that when the material thickness error is within the range of ±0.005 mm, the microwave absorption performance error of the multilayer absorber is about 5%. Within this error range, the performance of the multilayer absorber can be guaranteed. The sensitivity analysis results of the materials in each layer of the absorber indicate that the concentration and thickness of the graphite layer have the greatest impact on the performance of the absorber.
COMMUNICATIONS
Enhancing Network Throughput via the Equal Interval Frame Aggregation Scheme for IEEE 802.11ax WLANs
ZHU Yihua, XU Mengying
2023, 32(4): 747-759. doi: 10.23919/cje.2022.00.282
Abstract(323) HTML (161) PDF(36)
Abstract:
Frame aggregation is fully supported in the newly published IEEE 802.11ax standard to improve throughput. With frame aggregation, a mobile station combines multiple subframes into an aggregate MAC service data unit (A-MSDU) or an aggregate MAC protocol data unit (A-MPDU) for transmission. It is challenging for a mobile station in 802.11ax WLANs to set an appropriate number of subframes being included in an A-MSDU or A-MPDU. This problem is solved in this paper by the proposed equal interval frame aggregation (EIFA) scheme which lets a mobile station aggregate at most k subframes at a fixed time period of T. A novel Markov model is developed for deriving the probability of number of subframes in the data buffer at the mobile station, resulting in the throughput and packet delay in the EIFA. Moreover, the optimization problem of maximizing the throughput with the constraint on delay is formulated, and its solution leads to the optimal pair of parameters k and T for improving throughput in the EIFA scheme. Simulation results show the EIFA has a higher throughput than the ones in which the mobile station chooses the minimum, the maximum, or a random number of subframes.
On UAV Serving Node Deployment for Temporary Coverage in Forest Environment: A Hierarchical Deep Reinforcement Learning Approach
WANG Li, WU Xuewei, WANG Yanhui, XIAO Zhe, LI Liang, FEI Aiguo
2023, 32(4): 760-772. doi: 10.23919/cje.2021.00.326
Abstract(681) HTML (338) PDF(78)
Abstract:
Unmanned aerial vehicles (UAVs) can be effectively used as serving stations in emergency communications because of their free movements, strong flexibility, and dynamic coverage. In this paper, we propose a coordinated multiple points based UAV deployment framework to improve system average ergodic rate, by using the fuzzy C-means algorithm to cluster the ground users and considering exclusive forest channel models for the two cases, i.e., associated with a broken base station or an available base station. In addition, we derive the upper bound of the average ergodic rate to reduce computational complexity. Since deep reinforcement learning (DRL) can deal with the complex forest environment while the large action and state space of UAVs leads to slow convergence, we use a ratio cut method to divide UAVs into groups and propose a hierarchical clustering DRL (HC-DRL) approach with quick convergence to optimize the UAV deployment. Simulation results show that the proposed framework can effectively reduce the complexity, and outperforms the counterparts in accelerating the convergence speed.
Multi-Frequency-Ranging Positioning Algorithm for 5G OFDM Communication Systems
LI Wengang, XU Yaqin, ZHANG Chenmeng, TIAN Yiheng, LIU Mohan, HUANG Jun
2023, 32(4): 773-784. doi: 10.23919/cje.2021.00.124
Abstract(577) HTML (293) PDF(50)
Abstract:
The accurate determination of vehicle location is of great research significance, considering challenges such as the multipath environment and the absence of Global Navigation Satellite System (GNSS) signals. In this particular environment, vehicles equipped with 5G wireless communication devices can enhance their positioning accuracy by exchanging information with infrastructure (vehicle-to-infrastructure, V2I). Therefore, in this paper, we propose a multifrequency ranging method and positioning algorithm specifically designed for 5G orthogonal frequency division multiplexing (OFDM) communication systems. Our approach involves selecting specific subcarriers within the OFDM communication system for transmitting ranging frames and capturing delay observations. Importantly, this selection does not affect the functionality of other subcarriers used for regular communication. By utilizing dedicated subcarriers for ranging and positioning, we achieve accurate vehicle location without significantly impacting communication capacity. We outline the method for selecting ranging subcarriers and describe the format of the ranging frame carried by these subcarriers. To evaluate the effectiveness of our system, we prove the Cramér-Rao lower bound of this ranging positioning system. The obtained ranging positioning accuracy meets the requirements for vehicle location applications. In our experimental simulations, we compare the performance of our system with other positioning methods, demonstrating its superiority. Additionally, we provide theoretical proofs and simulations that establish the relationship between ranging accuracy and channel parameters in a multipath environment. The simulation results indicate that, under the conditions of a 5 GHz frequency and a high signal-to-noise ratio, our system achieves a positioning accuracy of approximately 5 cm.
Recursive Feature Elimination Based Feature Selection in Modulation Classification for MIMO Systems
ZHOU Shuai, LI Tao, LI Yongzhao
2023, 32(4): 785-792. doi: 10.23919/cje.2021.00.347
Abstract(637) HTML (313) PDF(45)
Abstract:
The feature-based (FB) algorithms are widely used in modulation classification due to their low complexity. As a prerequisite step of FB, feature selection can reduce the computational complexity without significant performance loss. In this paper, according to the linear separability of cumulant features, the hyperplane of the support vector machine is used to classify modulation types, and the contribution of different features is ranked through the weight vector. Then, cumulant features are selected using recursive feature elimination (RFE) to identify the modulation type employed at the transmitter. We compare the performance of the proposed algorithm with existing feature selection algorithms and analyze the complexity of all the mentioned algorithms. Simulation results verify that the proposed RFE algorithm can optimize the selection of the features to realize modulation recognition and improve identification efficiency.
Mobility Prediction Based Tracking of Moving Objects in Wireless Sensor Networks
TANG Chao, XIA Yinqiu, DOU Lihua
2023, 32(4): 793-805. doi: 10.23919/cje.2021.00.365
Abstract(566) HTML (282) PDF(43)
Abstract:
This paper investigates the multi-sensor fused localization of moving targets in a wireless sensor network. Each ultra-wide band (UWB) sensor is assigned a stability weight according to its survival time prediction. The measurement accuracy of each sensor into the constraints of the weight distribution based on the interactive multi-model method, a double weight distribution algorithm that considers measurement accuracy and stability is proposed. Based on the double weight algorithm, the measurement information of each UWB sensor, the inertial measurement unit (IMU)-based state vector and the UWB-based state vector by federated Kalman filter are integrated to realize the correction of the IMU. Finally, several numerical simulations are performed to show that the proposed algorithm can effectively suppress the measurement dropout when tracking moving targets in a wireless sensor network, and it can also automatically adjust the weight of each sensor based on the measurement error covariance to improve the tracking accuracy.
Asymptotically Optimal Golay-ZCZ Sequence Sets with Flexible Length
GU Zhi, ZHOU Zhengchun, Adhikary Avik Ranjan, FENG Yanghe, FAN Pingzhi
2023, 32(4): 806-820. doi: 10.23919/cje.2022.00.266
Abstract(489) HTML (242) PDF(30)
Abstract:
Zero correlation zone (ZCZ) sequences and Golay complementary sequences are two kinds of sequences with different preferable correlation properties. Golay-ZCZ sequences are special kinds of complementary sequences which also possess a large ZCZ and are good candidates for pilots in OFDM systems. Known Golay-ZCZ sequences reported in the literature have a limitation in the length which is the form of a power of 2. In this paper, we propose two constructions of Golay-ZCZ sequence sets with new parameters which generalize the constructions of Gong et al. (IEEE Trans. Commun., 61(9), 2013) and Chen et al. (IEEE Trans. Commun., 66(11), 2018). Notably, one of the two constructions generates optimal binary Golay-ZCZ sequences, while the other generates asymptotically optimal polyphase Golay-ZCZ sequences as the number of sequences increases. We also show, through numerical simulations, the applicability of the proposed Golay-ZCZ sequences in inter-symbol interference channel estimation. Interestingly, in certain application scenarios, the proposed Golay-ZCZ sequences performs better as compared to the existing state-of-the-art sequences.
Dispersed Computing Resource Discovery Model and Algorithm for Polymorphic Migration Network Architecture
ZHOU Chengcheng, ZHANG Lukai, ZENG Guangping, LIN Fuhong
2023, 32(4): 821-839. doi: 10.23919/cje.2022.00.305
Abstract(337) HTML (170) PDF(25)
Abstract:
Dynamic resource discovery in a network of dispersed computing resources is an open problem. The establishment and maintenance of resource pool information are critical, which involves both the polymorphic migration of the network and the time and energy costs resulting from node selection and frequent interactions of information between nodes. The resource discovery problem for dispersed computing can be considered a dynamic multi-level decision problem. A bi-level programming model of dispersed computing resource discovery is developed, which is driven by time cost, energy consumption and accuracy of information acquisition. The upper-level model is to design a reasonable network structure of resource discovery, and the lower-level model is to explore an effective discovery mode. Complex network topology features are used for the first time to analyze the polymorphic migration characteristics of resource discovery networks. We propose an integrated calibration method for energy consumption parameters based on two discovery modes (i.e., agent mode and self-directed mode). A symmetric trust region based heuristic algorithm is proposed for solving the system model. The numerical simulation is performed in a dispersed computing network with multiple modes and topological states, which proves the feasibility of the model and the effectiveness of the algorithm.
ARTIFICIAL INTELLIGENCE
Adaptive Tensor Rank Approximation for Multi-View Subspace Clustering
SUN Xiaoli, HAI Yang, ZHANG Xiujun, XU Chen
2023, 32(4): 840-853. doi: 10.23919/cje.2022.00.180
Abstract(571) HTML (284) PDF(53)
Abstract:
Multi-view subspace clustering under a tensor framework remains a challenging problem, which can be potentially applied to image classification, impainting, denoising, etc. There are some existing tensor-based multi-view subspace clustering models mainly making use of the consistency in different views through tensor nuclear norm (TNN). The diversity which means the intrinsic difference in individual view is always ignored. In this paper, a new tensorial multi-view subspace clustering model is proposed, which jointly exploits both the consistency and diversity in each view. The view representation is decomposed into view-consistent part (low-rank part) and view-specific part (diverse part). A tensor adaptive log-determinant regularization (TALR) is imposed on the low-rank part to better relax the tensor multi-rank, and a view-specific sparsity regularization is applied on the diverse part to ensure connectedness property. Although the TALR minimization is not convex, it has a closed-form analytical solution and its convergency is validated mathematically. Extensive evaluations on six widely used clustering datasets are executed and our model is demonstrated to have the superior performance.
A Model for Chinese Named Entity Recognition Based on Global Pointer and Adversarial Learning
ZHANG Yangsen, LI Jianlong, XIN Yonghui, ZHAO Xiquan, LIU Yang
2023, 32(4): 854-867. doi: 10.23919/cje.2022.00.279
Abstract(478) HTML (237) PDF(55)
Abstract:
To solve the problem that the Chinese named entity recognition (NER) models have poor anti-interference ability and inaccurate entity boundary recognition, this paper proposes the RGP-with-FGM model which is based on global pointer and adversarial learning. Firstly, the RoBERTa-WWM model is used to optimize the semantic representation of the text, and fast gradient method is used to add perturbation to the word embedding layer to enhance the robustness of the model. Then, BiGRU is used to focus on the timing information of Chinese characters to enhance the semantic connection. Finally, the global pointer is constructed in the decoding layer to obtain more accurate entity boundary recognition results. In order to verify the effectiveness of the model proposed in this paper, we construct Uyghur names dataset (UHND) to train the Chinese NER model, and perform extensive experiments with public Chinese NER data sets. Experimental results show that for UHND, the F1 value is 95.12%, which is 3.09% higher than that of the RoBERTa-WWM-BiGRU-CRF model. For the Resume data set, the Precision and F1 value are 96.28% and 96.10% respectively.
Deep Contextual Representation Learning for Identifying Essential Proteins via Integrating Multisource Protein Features
LI Weihua, LIU Wenyang, GUO Yanbu, WANG Bingyi, QING Hua
2023, 32(4): 868-881. doi: 10.23919/cje.2022.00.053
Abstract(1090) HTML (546) PDF(45)
Abstract:
Essential proteins with biological functions are necessary for the survival of organisms. Computational recognition methods of essential proteins can reduce the workload and provide candidate proteins for biologists. However, existing methods fail to efficiently identify essential proteins, and generally do not fully use amino acid sequence information to improve the performance of essential protein recognition. In this work, we propose an end-to-end deep contextual representation learning framework called DeepIEP to automatically learn biological discriminative features without prior knowledge based on protein network heterogeneous information. Specifically, the model attaches amino acid sequences as the attributes of each protein node in the protein interaction network, and then automatically learns topological features from protein interaction networks by graph embedding algorithms. Next, multi-scale convolutions and gated recurrent unit networks are used to extract contextual features from gene expression profiles. The extensive experiments confirm that our DeepIEP is an effective and efficient feature learning framework for identifying essential proteins and contextual features of protein sequences can improve the recognition performance of essential proteins.
Teacher-Student Training Approach Using an Adaptive Gain Mask for LSTM-Based Speech Enhancement in the Airborne Noise Environment
HUANG Ping, WU Yafeng
2023, 32(4): 882-895. doi: 10.23919/cje.2022.00.307
Abstract(250) HTML (128) PDF(18)
Abstract:
Research on speech enhancement algorithms in the airborne environment is of great significance to the security of airborne systems. Recently, the research focus of speech enhancement has turned from conventional unsupervised algorithms, like the log minimum mean square error estimator (log-MMSE), to the state-of-the-art masking-based long short-term memory (LSTM) method. However, each method has its characteristics and limitations, so they cannot always handle noise well. Besides, the requirements of clean speech and noise data for training a supervised speech enhancement model are difficult to satisfy in the real-world airborne environment. Therefore, in this work, to fully utilize the advantages of those two different methods without any data restrictions, we propose a novel adaptive gain mask (AGM) based teacher-student training approach for speech enhancement. In our method, the AGM, as a robust learning target for the student model, is devised by incorporating the estimated ideal ratio mask from the teacher model into the procedure of the log-MMSE approach. To get an appropriate tradeoff between the two methods, we adaptively update the AGM using a recursive weighting coefficient. Experiments on the real airborne data show that the proposed AGM-based method outperforms other baselines in terms of all essential objective metrics evaluated in this paper.
Attention Guided Enhancement Network for Weakly Supervised Semantic Segmentation
ZHANG Zhe, WANG Bilin, YU Zhezhou, ZHAO Fengzhi
2023, 32(4): 896-907. doi: 10.23919/cje.2021.00.230
Abstract(638) HTML (328) PDF(43)
Abstract:
Weakly supervised semantic segmentation using only image-level labels is critical since it alleviates the need for expensive pixel-level labels. Most cutting-edge methods adopt two-step solutions that learn to produce pseudo-ground-truth using only image-level labels and then train off-the-shelf fully supervised semantic segmentation network with these pseudo labels. Although these methods have made significant progress, they also increase the complexity of the model and training. In this paper, we propose a one-step approach for weakly supervised image semantic segmentation—attention guided enhancement network (AGEN), which produces pseudo-pixel-level labels under the supervision of image-level labels and trains the network to generate segmentation masks in an end-to-end manner. Particularly, we employ class activation maps (CAM) produced by different layers of the classification branch to guide the segmentation branch to learn spatial and semantic information. However, the CAM produced by the lower layer can capture the complete object region but with many noises. Thus, the self-attention module is proposed to enhance object regions adaptively and suppress irrelevant object regions, further boosting the segmentation performance. Experiments on the Pascal VOC 2012 dataset demonstrate that AGEN outperforms alternative state-of-the-art weakly supervised semantic segmentation methods exclusively relying on image-level labels.
Colour Variation Minimization Retinex Decomposition and Enhancement with a Multi-Branch Decomposition Network
DENG Jiawei, YU Zhenming, PANG Guangyao
2023, 32(4): 908-919. doi: 10.23919/cje.2021.00.350
Abstract(243) HTML (120) PDF(17)
Abstract:
This paper proposes a colour variation minimization retinex decomposition and enhancement with a multi-branch decomposition network (CvmD-net) to remove single image darkness. The network overcomes the problem that retinex deep learning model relies on matching bright images to process dark images. Specifically, our method takes two stages to light up the darkness in initial images: image decomposition and brightness optimization. We propose an input constant feature prior mechanism (ICFP) based on reflection constant features. The mechanism extracts structure and colour from the input images and constrains the reflected images output from the decomposition model to reduce color distortion and artifacts. The noise amplification during decomposition is addressed by a multi-branch decomposition network. Sub-networks with different structures are employed to focus on different prediction tasks. This paper proposes a reference mechanism for input brightness. This mechanism optimizes the output brightness distribution by calculating the reference brightness of the dark images. Experimental results on two benchmark datasets, namely, LOL and ZeroDCE, demonstrate that the proposed method can better balance dense noise interference and colour restoration. For the evaluation on real images, we collect Skynet images at night to verify the performance of the proposed approach. Compared with the state-of-the-art non-reference retinex decomposition-enhancement models, this paper has the best brightness optimization.
Self-Adaptive Discrete Cuckoo Search Algorithm for the Service Routing Problem with Time Windows and Stochastic Service Time
ZHANG Guoyun, WU Meng, LI Wujing, OU Xianfeng, XIE Wenwu
2023, 32(4): 920-931. doi: 10.23919/cje.2022.00.072
Abstract(297) HTML (149) PDF(28)
Abstract:
Making house calls is very crucial to deal with the competitive pressures of the service business and to improve service quality. We design a model called service routing problem with time windows and stochastic service time (SRPTW-SST) that is based on vehicle routing problem with time windows. A self-adaptive discrete cuckoo search algorithm with genetic mechanism (sDCS-GM) is proposed for the model SRPTW-SST. Moreover, we design a selection mechanism to improve the logicality of the algorithm based on the strong randomness of the Lévy flight. We introduce a genetic mechanism and design a neighborhood search mechanism for improving the robustness of the algorithm. In addition, an adaptive parameter adjustment method is designed to eliminate the impact of fixed parameters. The experimental results show that the sDCS-GM algorithm is more robust and effective than the state-of-the-art methods.
Rating Prediction Model Based on Causal Inference Debiasing Method in Recommendation
NAN Jiangang, WANG Yajun, WANG Chengcheng
2023, 32(4): 932-940. doi: 10.23919/cje.2022.00.076
Abstract(391) HTML (196) PDF(39)
Abstract:
The rating prediction task plays an important role in the recommendation model. Most existing methods predict ratings by extracting user and items characteristics from historical review data. However, the recommended strategies in historical review data are often based on partial observational data, which having the problems of unbalanced distribution, lack of robustness, and inability to obtain unbiased prediction results. Therefore, a novel rating prediction model based on causal inference debiasing (CID) method is proposed. The model can mitigate the negative effects of context bias and improve the robustness by studying the causal relationship between review information and user ratings. The proposed CID rating prediction model is plug-and-play and is not limited to one baseline prediction method. The proposed method is tested on four open datasets. The results show that the proposed method is feasible. Compared with the most advanced models, the prediction accuracy of the CID rating prediction model has been further improved. The experimental results show the debiasing effectiveness of the CID rating prediction model.
Corrigenda
2023, 32(4): 941-941.
Abstract(46) HTML (25) PDF(89)
Abstract: