通讯机构:
[Zhang, C.] W;Wuhan Polytechnic University, China
关键词:
Preload information;scattering transform;feature granularity consistency;self-attention mechanism;focal loss
摘要:
Urban sound event detection can automatically preload relevant information for a robot to ensure that it can be applied to various scene-activity tasks. To address the limitations of timbre similarity and scene recognition by audio collection devices, a fusion model based on the self-attention mechanism is proposed in this paper. The model consists of scattering transform and self-attention model. The scattering transform computes modulation spectrum coefficients of multiple orders through cascades of wavelet convolutions and modulus operators. It is learnable compared with Mel-scale Frequency Cepstral Coefficients (MFCC), and can be used to better restore the semantic features of some sound scenes with similar timbres. The transformer has an outstanding effect on Natural Language Processing (NLP) owing to its self-attention mechanism. In this paper, the self-attention mechanism in its encoder was used in the model, mainly to make the feature granularity consistent to refine the features. In addition, Focal Loss function was adopted in the model to curb the sample distribution imbalance. The Google Command and ESC-50 were used to supplement the scene categories of dataset UrbanSound8K. The model parameters of the learnable filters that performed well on the dataset UrbanSound8K were preserved to fine-tune the other two datasets with insufficient data volume and more target categories. The length of slice duration was further explored the in the model. The experimental results show that the model can achieve better performance in a large range of scene models.
会议名称:
IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) / IEEE World Congress on Computational Intelligence (IEEE WCCI) / International Joint Conference on Neural Networks (IJCNN) / IEEE Congress on Evolutionary Computation (IEEE CEC)
会议时间:
JUL 18-23, 2022
会议地点:
Padua, ITALY
会议主办单位:
[Bai, Jun;Sajjanhar, Atul;Xiang, Yong] Deakin Univ, Sch Informat Technol, Geelong, Vic, Australia.^[Tong, Xiaojun] Wuhan Text Univ, Sch Comp Sci & Artificial Intelligence, Wuhan, Peoples R China.^[Zeng, Shan] Wuhan Polytech Univ, Sch Math & Comp Sci, Wuhan, Peoples R China.
会议论文集名称:
IEEE International Joint Conference on Neural Networks (IJCNN)
关键词:
distributed learning;federated learning;data heterogeneity;Non-IID data;heterogeneous model fusion
摘要:
Federated Learning (FL) offers a novel distributed machine learning context whereby a global model is collaboratively learned through edge devices without violating data privacy. However, intrinsic data heterogeneity in the federated network can induce model heterogeneity, thus posing a great challenge to the server-side model aggregation performance. Existing FL algorithms widely adopt model-wise weighted averaging for client models to generate the new global model, which emphasizes the importance of the holistic model but ignores the importance of distinctions between internal parameters of various client models. In this paper, we propose a novel parameter-wise elastic weighted averaging aggregation approach to realize the rapid fusion of heterogeneous client models. Specifically, each client evaluates the importance of model internal parameters in the model update and obtains the corresponding parameter importance coefficient vector; the server implements the parameter-wise weighted averaging for each parameter based on their importance coefficient vectors, thereby aggregating a new global model. Extensive experiments on MNIST and CIFAR-10 datasets with diverse network architectures and hyper-parameter combinations show that our proposed algorithm outperforms the existing state-of-the-art FL algorithms on the performance of heterogeneous model fusion.
摘要:
The construction of quantum maximum distance separable (MDS for short) codes is one of the hot issues in quantum information theory. As far as we know, researchers have done a lot of constructive work in the construction of quantum MDS codes. However, the known results do not cover all parameters. In this paper, we propose an efficient construction implemented by concatenating two existing quantum MDS codes. Compared to a previous work (Fang and Luo in Quantum Inf Process 19(1):16, 2020), we relax the restrictions of the construction and propose some new quantum MDS codes.
摘要:
The crux of image deraining stems from the challenge of recognizing the diverse rain patterns within the rainy image. Most methods for image deraining remain visible rain residuals in the restored image, which suffers from insufficient modeling of rain streaks. In this work, we propose contrastive learning-based generative network (CLGNet), which follows a coarse-to-fine framework. In the coarse phase, our CLGNet employs the hierarchical encoder-decoder structure to remove obvious rain patterns, and first generates the coarse background image. Then, we introduce a well-designed multiscale feature aggregation module in the refining phase to extract and integrate global information dependencies from different scales. In additon, to facilitate the intra-stage and cross-stage information interaction, we propose the intra-stage feature fusion module and the cross-stage feature fusion module to encode broad contextual information. More importantly, we propose an innovative contrastive learning strategy and apply it to each stage of our proposed CLGNet to enhance the decoupling ability of the encoder and help the model recognize complex rain patterns. Extensive experiments on five benchmark datasets demonstrate the superiority of our proposed CLGNet than other state-of-the-art methods for single image deraining on both the visual quality and quantitative evaluation. (C) 2022 SPIE and IS&T
作者机构:
[Yayi Huang; Changhua Liu] School of Mathematics and Computer Science, Wuhan Polytechnic University, Wuhan, Hubei, 4300232, China;[Qiming Ma; Xiaoming Wu; Hao Li; Kun Xu; Gaoxiang Ji; Fang Qian; Lixia Li; Qian Huang; Ying Long; Xiaojun Zhang; Biyun Chen] Oil Crops Research Institute, Chinese Academy of Agricultural Sciences, Key Laboratory for Oil Crop Biology and Genetic Improvement, Ministry of Agriculture, Wuhan, Hubei Province, 430062, China
通讯机构:
[Biyun Chen] O;[Changhua Liu] S;Oil Crops Research Institute, Chinese Academy of Agricultural Sciences, Key Laboratory for Oil Crop Biology and Genetic Improvement, Ministry of Agriculture, Wuhan, Hubei Province, 430062, China<&wdkj&>School of Mathematics and Computer Science, Wuhan Polytechnic University, Wuhan, Hubei, 4300232, China
摘要:
The chlorophyll content has a direct effect on photosynthesis of crops. In order to explore a quick and convenient method for estimating the chlorophyll content of Brassica napus and facilitate efficient crop monitoring, we measured the actual value of chlorophyll with a SPAD-502 chlorophyll detector, and collected aerial images of B. napus with an unmanned aerial vehicle(UAV) carrying a RGB camera in this study. The total number of 270 samples collected images were divided into regions according to the planting conditions of different B. napus varieties in the field. Then, according to the empirical formula, there were 36 colors’ characteristic parameters calculated and combined. To estimate the chlorophyll content of rape, 189 samples were included in the modeling set, while the other 81 samples were enrolled in the validation set for testing the accuracy of this model. After the combination of R (red), G (green) and B (blue) color channels, the results showed that the color characteristics B/(R + G), b, B/G, (G-B)/(G + B), g-b were highly connected with the measured value of chlorophyll SPAD, and the correlation coefficient between the combination based on B/(R + G) and SPAD value was 0.747. With R2 = 0.805, RMSE = 3.343, and RE = 6.84%, the regression model created using random forest had superior outcomes, according to the model comparison. This study offers a new method for quickly estimating the amount of chlorophyll in rapeseed and a workable reference for crop monitoring using the UAV platform.
作者机构:
[曹艺怀; 吴礼发; 陈伟] School of Computer Science, Nanjing University of Posts and Telecommunications, Nanjing, 210023, China;[张帆] School of Mathematics and Computer Science, Wuhan Polytechnic University, Wuhan, 430048, China
通讯机构:
[Cao, Y.] S;School of Computer Science, China
关键词:
流量分析;异常检测
摘要:
WebShell是一种常见的Web脚本入侵工具.随着流量加密和代码混淆等技术的逐渐发展,使用传统的文本内容特征和网络流特征进行匹配的检测手段越来越难以防范生产环境下复杂的WebShell恶意攻击事件,特别是对于对抗性样本、变种样本或0Day漏洞样本的检测效果不够理想.搭建网络采集环境,在高速网络环境中利用数据平面开发套件(DPDK,data plane development kit)技术捕获网络数据包,标注了一套由1万余条不同平台、不同语言、不同工具、不同加密混淆方式的WebShell恶意流量与3万余条正常流量组成的数据集;通过异步流量分析系统框架和轻量型日志采集组件快速地解析原始流量,并融合专家知识深度分析几种流行的WebShell管理工具通信过程中的HTTP数据包,从而构建面向加密混淆型WebShell流量的有效特征集;基于该有效特征集使用支持向量机(SVM,support vector machine)算法实现对加密混淆型WebShell恶意流量的离线训练和在线检测.同时,利用遗传算法改进参数搜索方式,克服了由人工经验设置参数方位以及网格搜索陷入局部最优解的缺点,模型训练效率也得到提升.实验结果显示,在自建的WebShell攻击流量数据集上,保证了检测高效性的同时,检测模型的精确率为97.21%,召回率为98.01%,且在对抗性WebShell攻击的对比实验中表现良好.结果表明,所提方法能够显著降低WebShell攻击风险,可以对现有的安全监控体系进行有效补充,并在真实网络环境中部署和应用.
期刊:
2022 IEEE 13th International Conference on Software Engineering and Service Science (ICSESS),2022年:271-275
作者机构:
[Wangyang Shen; Weiping Jin] College of Food Science and Engineering, Wuhan Polytechnic University, Wuhan, China;[Qing Zhao] Xiangyang Tianyuan Lohas Rice Industry Co. Ltd, Xiangyang, China;[Guangbin Li] Qianjiang Jujin Rice Industry Co. Ltd, Qianjiang, China;[Liang Peng; Kang Zhou] College of Mathematics and Computer Science, Wuhan Polytechnic University, Wuhan, China
关键词:
Raw material index content prediction;Depth feature selection technology;Structure and parameter optimization of prediction model.
摘要:
In this paper, the depth feature selection technology is proposed. According to the different characteristics of feature extraction methods, a multi-layer feature extraction structure is formed, and selection variables are introduced to construct a global optimization model that enhances the feature expression of data sets from horizontal to vertical, so as to realize the adaptive feature selection of the prediction model as a whole. The real-coded harmony search algorithm was combined with BP, RNN and RBF neural network to optimize model structure and parameter. Experiments show that compared with the traditional prediction model, this method improves the prediction accuracy of each index value of raw materials corresponding to yellow rice wine products. The model determination coefficient is increased by 6.89%, and the mean square error is reduced by 7.43%. Food processing enterprises can select raw materials according to the predicted raw material index value.
作者机构:
[Wang, Zemin; An, Jiachun; Ma, Yuanyuan; Li, Fei; Liu, Shunlun; Li, Bing] Wuhan Univ, Chinese Antarctic Ctr Surveying & Mapping, Wuhan 430079, Peoples R China.;[Ma, Yuanyuan; Li, Fei] Wuhan Univ, Elect Informat Sch, Wuhan 430072, Peoples R China.;[Wang, Zemin; An, Jiachun; Ma, Yuanyuan; Li, Fei; Liu, Shunlun] MNR, Key Lab Polar Surveying & Mapping, Wuhan 430079, Peoples R China.;[Li, Bing] Wuhan Polytech Univ, Sch Math & Comp Sci, Wuhan 430048, Peoples R China.;[Ma, Weifeng] Yunnan Normal Univ, Fac Geog, Kunming 650050, Yunnan, Peoples R China.
通讯机构:
[Wang, ZM ] W;Wuhan Univ, Chinese Antarctic Ctr Surveying & Mapping, Wuhan 430079, Peoples R China.;MNR, Key Lab Polar Surveying & Mapping, Wuhan 430079, Peoples R China.
关键词:
glacier mass balance;SAR interferometry;ice motion;ionospheric effects;reformulation of the split-spectrum method;ionospheric correction
摘要:
Ice motion is an essential element for accurately evaluating glacier mass balance. Interferometric synthetic aperture radar (InSAR) has been widely applied for monitoring ice motion with high precision and wide coverage in the Antarctic. However, the ionospheric effects can significantly impact InSAR-based ice-motion measurements. At low radar frequencies in particular, the ionospheric effects have been regarded as a serious source of noise in L-band SAR data. The split-spectrum method (SSM) is commonly used for correcting the ionospheric effects of the InSAR technique. However, it requires spatial filtering with the relatively large factors used to scale the sub-bands' interferograms, which often results in an unwrapped phase error. In this paper, a reformulation of the split-spectrum method (RSSM) is introduced to correct the ionospheric effects in the Grove Mountains of East Antarctica, which have slow ice flow and frequent ionosphere changes. The results show that RSSM can effectively correct the ionospheric effects of InSAR-based ice-motion measurements. To evaluate the ability of ionospheric correction using RSSM, the result of ionospheric correction derived from SSM is compared with the results of RSSM. In addition, ionosphere-corrected ice motion is also compared with GPS and MEaSUREs. The results show that the ionosphere-corrected ice velocities are in good agreement with GPS observations and MEaSUREs. The average ice velocity from the InSAR time series is compared to that from MEaSUREs, and the average ionosphere-corrected ice velocity error reduces 43.9% in SSM and 51.1% in RSSM, respectively. The ionosphere-corrected ice velocity error is the most significant, reducing 86.9% in SSM and 90.4% in RSSM from 1 November 2007 to 19 December 2007. The results show that the ability of RSSM to correct ionospheric effects is slightly better than that of SSM. Therefore, we deduce that the RSSM offers a feasible way to correct ionospheric effects in InSAR-based ice-motion measurements in Antarctica.
摘要:
Hyperspectral unmixing (HU) is a fundamental and critical task in various hyperspectral image (HSI) applications. Over the past few years, the linear mixing model (LMM) has received widely attention for its high efficiency, definite physical meaning, and being amenable to mathematical treatment. Among the various linear unmixing methods, the autoencoder unmixing network has achieved superior performance and presented more significant potential because of the powerful data fitting ability and deep feature acquisition. However, the autoencoder unmixing network, focusing on the pixel-level associations, ignores the overall distribution and long-range dependencies of materials. Inspired by the receptive field mechanism and the effectiveness of multi-stage framework, we propose a multi-stage convolutional autoencoder network for hyperspectral linear unmixing, called MSNet. MSNet is capable of learning broad contextual information without losing the detailed features by the progressively multi-stage unmixing network in the unmixing process. Compared with the conventional single-stage unmixing methods, the multi-stage framework is more robust in solving the ill-posed unmixing problem. The proposed MSNet performs more effectively and competitively than state-of-the-art algorithms by comparison experiments on synthetic and real hyperspectral datasets. The source code is available at https://github.com/yuyang95/JAG-MSNet.
摘要:
Abstract: Erdős–Faber–Lovász conjecture states that if a graph G is a union of the n edge-disjoint copies of complete graph K n , that is, each pair of complete graphs has at most one shared vertex, then the chromatic number of graph G is n. In fact, we only need to consider the graphs where each pair of complete graphs has exactly one shared vertex. However, each shared vertex may be shared by more than two complete graphs. Therefore, this paper first considers the graphs where each shared vertex happens to be shared by two complete graphs, and then discusses the graphs with only one shared vertex shared by more than two complete graphs. The conjecture is correct for these two kinds of graphs in this work. Finally, the graph where each shared vertex happens to be shared by three complete graphs has been studied, and the conjecture also holds for such graphs when n = 13 . The graphs discussed in this paper have certain symmetric properties. The symmetry of graphs plays an important role in coloring. This work is an attempt to combine the symmetry of graphs with the coloring of graphs. Keywords: Erdős–Faber–Lovász conjecture; complete graph; shared vertex; hypergraph; coloring
期刊:
Cryptography and Communications,2022年14(1):145-159 ISSN:1936-2447
通讯作者:
Luo, Jinquan(luojinquan@mail.ccnu.edu.cn)
作者机构:
[Fang, Xiaolei] School of Mathematics and Computer Science, Wuhan Polytechnic University, Wuhan, 430048, China;[Jin, Renjie] College of Liberal Arts and Sciences, National University of Defense Technology, Changsha, 410073, China;[Ma, Wen; Luo, Jinquan] School of Mathematics and Statistics & Hubei Key Laboratory of Mathematical Sciences, Central China Normal University, Wuhan, 430079, China
通讯机构:
[Jinquan Luo] S;School of Mathematics and Statistics & Hubei Key Laboratory of Mathematical Sciences, Central China Normal University, Wuhan, China
摘要:
Clustering noisy, high-dimensional, and structurally complex data has always been a challenging task. As most existing clustering methods are not able to deal with both the adverse impact of noisy samples and the complex structures of data, in this paper, we propose a novel Robust and Sparse Possibilistic K-Subspace Clustering algorithm (RSPKS) to integrate subspace recovery and possibilistic clustering algorithms under a unified sparse framework. First, the proposed method sparsifies the membership matrix and the subspace projection vector under a dual-sparse framework to handle high-dimensional noisy data. This unifies dimensionality reduction and clustering using one objective function, for which the optimization can be realized through synchronous iteration. Second, the reconstruction error of each sample in the local subspace is used as the distance metric for classification. That is, each sample itself is treated as a clustering prototype so as not to be affected by the structure of the overall data distribution. Therefore, the clustering prototype construction problem of data with complex structures can be better addressed. Finally, to deal with non-linear regions, our RSPKS method is further extended into a kernelized version, namely the Kernelized Robust and Sparse Possibilistic K-Subspace Clustering (KRSPKS) algorithm. Experimental results on both synthetic and real-world datasets demonstrate that our proposed method outperforms state-of-the-art algorithms in terms of clustering accuracy. IEEE