Deep convolutional neural network has strong feature extraction and fitting capabilities and perform well in hyperspectral image classification tasks. However, due to its huge parameters, complex structure and high energy consumption, it is difficult to be used in mobile edge computing. Spiking neural network (SNN) has the characteristics of event-driven and low energy consumption and has developed rapidly in image classification. But it usually requires more time steps to achieve optimal accuracy. This paper designs a faster residual multi-branch SNN (FRM-SNN) based on leaky integrate-and-fire neurons for HSI classification. The network uses the residual multi-branch module (RMM) as the basic unit for feature extraction. The RMM is composed of spiking mixed convolution and spiking point convolution, which can effectively extract spatial spectral features. Secondly, to address the problem of non-differentiability of Dirac function spiking propagation, a simple and efficient arcsine approximate derivative was designed for gradient proxy, and the classification performance, testing time, and training time of various approximate derivative algorithms were analyzed and evaluated under the same network architecture. Experimental results on six public HSI data sets show that compared with advanced SNN-based HSI classification algorithms, the time step, training time and testing time required for FRM-SNN to achieve optimal accuracy are shortened by approximately 84%, 63% and 70%. This study has important practical significance for promoting the engineering application of HSI classification algorithms in unmanned autonomous devices such as spaceborne and airborne systems.

Extreme rainfall can severely affect all vegetation types, significantly impacting crop yield and quality. This study aimed to assess the response and recovery of vegetation phenology to an extreme rainfall event (with total weekly rainfall exceeding 500 mm in several cities) in Henan Province, China, in 2021. The analysis utilized multi-sourced data, including remote sensing reflectance, meteorological, and crop yield data. First, the Normalized Difference Vegetation Index (NDVI) time series was calculated from reflectance data on the Google Earth Engine (GEE) platform. Next, the ‘phenofit’ R language package was used to extract the phenology parameters—the start of the growing season (SOS) and the end of the growing season (EOS). Finally, the Statistical Package for the Social Sciences (SPSS, v.26.0.0.0) software was used for Duncan’s analysis, and Matrix Laboratory (MATLAB, v.R2022b) software was used to analyze the effects of rainfall on land surface phenology (LSP) and crop yield. The results showed the following. (1) The extreme rainfall event’s impact on phenology manifested directly as a delay in EOS in the year of the event. In 2021, the EOS of the second growing season was delayed by 4.97 days for cropland, 15.54 days for forest, 13.06 days for grassland, and 12.49 days for shrubland. (2) Resistance was weak in 2021, but recovery reached in most areas by 2022 and slowed in 2023. (3) In each year, SOS was predominantly negatively correlated with total rainfall in July (64% of cropland area in the first growing season, 53% of grassland area, and 71% of shrubland area). In contrast, the EOS was predominantly positively correlated with rainfall (51% and 54% area of cropland in the first and second growing season, respectively, and 76% of shrubland area); however, crop yields were mainly negatively correlated with rainfall (71% for corn, 60% for beans) and decreased during the year of the event, with negative correlation coefficients between rainfall and yield (−0.02 for corn, −0.25 for beans). This work highlights the sensitivity of crops to extreme rainfall and underscores the need for further research on their long-term recovery.

In recent years, the analysis of legal judgments and the prediction of outcomes based on case factual descriptions have become hot research topics in the field of judiciary. Among them, the task of charge prediction aims to predict the applicable charges of a judicial case based on its factual description, making it an important research area in the intelligent judiciary. While significant progress has been made in machine learning and deep learning, traditional methods are limited to handling data in Euclidean space and cannot effectively capture the semantic information in the text. To overcome the limitations of traditional learning approaches, many studies have started exploring the use of graphs to represent rich relationships between entities in text and employing graph convolutional neural networks to learn text representations. In this paper, we propose a charge prediction method based on graph convolutional neural networks. By constructing a similarity graph between cases and utilizing graph convolutional neural networks to learn case feature representations, we can better capture the relational information between cases and improve the accuracy of charge prediction. Experimental results on multiple benchmark datasets demonstrate that our proposed model outperforms traditional methods in charge prediction tasks.

Side-scan sonar (SSS) images have a wide range of applications in underwater target detection and recognition. However, due to the complexity of the underwater environment, the classification performance of sonar images is usually constrained by issues such as noise and inconspicuous texture features, including speckle noise, sensor noise, and interference from other sources. These noises can degrade the image quality, making it challenging to extract meaningful features and affecting the performance of classification algorithms. To address these challenges, we propose a novel classification model named Shuffle-RDSNet for SSS images. Specifically, we design the residual dual-path shrinkage network (RDSNet), which utilizes a soft thresholding function and determines the threshold by combining two paths to extract features from varying scales. The RDSNet is then integrated with the ShuffleNet V2 network to construct the proposed Shuffle-RDSNet model. This approach effectively mitigates the effect of noise in the feature extraction process and enhances the classification performance of the model. Furthermore, we employ additional techniques, including dilated convolution and depthwise separable convolution (DSC), to optimize the model and further enhance the classification accuracy and performance of the proposed method. Experimental results show that our model outperforms other classification models with a classification accuracy of up to 96.74

Air pollution is an important issue affecting sustainable development in China, and accurate air quality prediction has become an important means of air pollution control. At present, traditional methods, such as deterministic and statistical approaches, have large prediction errors and cannot provide effective information to prevent the negative effects of air pollution. Therefore, few existing methods could obtain accurate air pollutant time series predictions. To this end, a deep learning-based air pollutant prediction method, namely, the autocorrelation error-Informer (AE-Informer) model, is proposed in this study. The model implements the AE based on the Informer model. The AE-Informer model is used to predict the hourly concentrations of multiple air pollutants, including PM 10 , PM 2.5 , NO 2 , and O 3 . The experimental results show that the mean absolute error (MAE) and root mean square error (RMSE) values of AE-Informer in multivariate prediction are 3% less than those of the Informer model; thus, the prediction error is effectively reduced. In addition, a stacking ensemble model is proposed to supplement the missing air pollutant time series data. This study uses Henan Province in China as an example to test the validity of the proposed methodology.

Feature selection for high-dimensional data is an important issue in machine learning, pattern recognition and bioinformatics fields. Feature selection algorithms are proposed to select the relevant feature subset from the original features. To adaptively identify the important highly correlated features from high-dimensional data which often beneficial to improve classification accuracy is a challenge. In this paper, we propose a regularized logistic regression with adaptive Lasso and correlation based penalty model to select informative highly correlated features adaptively. To incorporate significance of features into regression model, we first measure significance of each feature based on mutual information, and propose an adaptive weight construction strategy. Based on the adaptive weight construction strategy, the proposed adaptive logistic regression can impose a large amount of penalty on irrelevant features, and thus noise features are easily removed from the model and remain the informative features. The experimental results on the simulation and realworld datasets demonstrate the effectiveness and the superiority the proposed model by comparing it to existing competing regularized logistic regression models.(c) 2021 Elsevier Inc. All rights reserved.

With the acceleration of urbanization, ozone (O3) pollution has become increasingly serious in many Chinese cities. This study analyzes the temporal and spatial characteristics of O3 based on monitoring and meteorological data for 366 cities and national weather stations throughout China from 2016 to 2020. Least squares linear regression and Spearman’s correlation coefficient were computed to investigate the relationships of O3 with various pollution factors and meteorological conditions. Global Moran’s I and the Getis–Ord index Gi* were adopted to reveal the spatial agglomeration of O3 pollution in Chinese cities and characterize the temporal and spatial characteristics of hot and cold spots. The results show that the national proportion of cities with an annual concentration exceeding 160 μg·m−3 increased from 21.6% in 2016 to 50.9% in 2018 but dropped to 21.5% in 2020; these cities are concentrated mainly in Central China (CC) and East China (EC). Throughout most of China, the highest seasonal O3 concentrations occur in summer, while the highest values in South China (SC) and Southwest China (SWC) occur in autumn and spring, respectively. The highest monthly O3 concentration reached 200 μg·m−3 in North China (NC) in June, while the lowest value was 60 μg·m−3 in Northeast China (NEC) in December. O3 is positively correlated with the ground surface temperature (GST) and sunshine duration (SSD) and negatively correlated with pressure (PRS) and relative humidity (RHU). Wind speed (WIN) and precipitation (PRE) were positively correlated in all regions except SC. O3 concentrations are significantly differentiated in space: O3 pollution is high in CC and EC and relatively low in the western and northeastern regions. The concentration of O3 exhibits obvious agglomeration characteristics, with hot spots being concentrated mainly in NC, CC and EC.

The underwater environment is complicated and changeable and contains many noises, making it difficult to detect a particular object in the underwater environment. At present, the main seabed detection technology explores the seabed environment with sonar equipment. However, the characteristics of underwater sonar imaging (e.g., low contrast, blurred edges, poor texture, and unsatisfactory quality) have serious negative influences on such image classification. Therefore, in this study, we propose a dual-path deep residual “shrinkage” network (DP-DRSN) module, which is a simple and effective neural network attention module that can classify side-scan sonar images. Specifically, the module can extract background and feature texture information of the input feature mapping through different scales (e.g., global average pooling and global max pooling), whereas scale information passes through a two-layer 1 × 1 convolution to increase nonlinearity. This helps realize cross-channel information interaction and information integration simultaneously before outputting threshold parameters in a sigmoid layer. The parameters are then multiplied by the average value of the input feature mapping to obtain a threshold, which is used to denoise the image features using the soft threshold function. The proposed DP-DRSN study provided higher classification accuracy and efficiency than other models. In this way, the feasibility and effectiveness of DP-DRSN in image classification of side-scan sonar are proven.

针对我国研究生教育中存在培养模式单一、学生动手实践能力提升不明显的问题,提出高校与企业共建研究生教育创新培养基地的新模式,以河南大学—中科空间信息(廊坊)研究院研究生教育创新培养基地为例进行探索,结果表明共建研究生教育创新培养基地可以有效提高工科研究生的创新能力和解决实际问题的能力.

基于2001～2018年MODIS标准产品,研究了我国及七大区域热异常点的时空分布特征.结果表明:空间分布上,热异常点主要分布在除西北、西南之外的大部分地区;年际趋势上,2001～2014年间热异常点数量持续上升,年均增长率为15.01％,2015年后逐年下降,年均下降率为14.96％.月季尺度上,热异常点在春、秋季节出现最为频繁(春:551 716个,秋:416 698个),春、秋季相对在东北地区分布最多(春:164 898个,秋:186 727个),东北地区月均数量10月最高(118 274个);夏季热异常点数量最低(290 793个),多分布于华东地区(120 455个),华东地区月均数量6月最高(76 465个);冬季数量为358 483个,且在华南地区分布最多(108 209个),华南地区月均数量1月最高(37 770个).研究有助于掌握我国典型区域的森林、草原火灾,以及由于秸秆焚烧、工业排放等引起热异常的变化情况,进而为区域灾害防治和环境监测提供技术支撑.

Most deep learning methods in hyperspectral image (HSI) classification use local learning methods, where overlapping areas between pixels can lead to spatial redundancy and higher computational cost. This paper proposes an efficient global learning (EGL) framework for HSI classification. The EGL framework was composed of universal global random stratification (UGSS) sampling strategy and a classification model BrsNet. The UGSS sampling strategy was used to solve the problem of insufficient gradient variance resulted from limited training samples. To fully extract and explore the most distinguishing feature representation, we used the modified linear bottleneck structure with spectral attention as a part of the BrsNet network to extract spectral spatial information. As a type of spectral attention, the shuffle spectral attention module screened important spectral features from the rich spectral information of HSI to improve the classification accuracy of the model. Meanwhile, we also designed a double branch structure in BrsNet that extracted more abundant spatial information from local and global perspectives to increase the performance of our classification framework. Experiments were conducted on three famous datasets, IP, PU, and SA. Compared with other classification methods, our proposed method produced competitive results in training time, while having a greater advantage in test time.

Side-scan sonar is widely used in underwater rescue and the detection of undersea targets, such as shipwrecks, aircraft crashes, etc. Automatic object classification plays an important role in the rescue process to reduce the workload of staff and subjective errors caused by visual fatigue. However, the application of automatic object classification in side-scan sonar images is still lacking, which is due to a lack of datasets and the small number of image samples containing specific target objects. Secondly, the real data of side-scan sonar images are unbalanced. Therefore, a side-scan sonar image classification method based on synthetic data and transfer learning is proposed in this paper. In this method, optical images are used as inputs and the style transfer network is employed to simulate the side-scan sonar image to generate "simulated side-scan sonar images"; meanwhile, a convolutional neural network pre-trained on ImageNet is introduced for classification. In this paper, we experimentally demonstrate that the maximum accuracy of target classification is up to 97.32% by fine-tuning the pre-trained convolutional neural network using a training set incorporating "simulated side-scan sonar images". The results show that the classification accuracy can be effectively improved by combining a pre-trained convolutional neural network and "similar side-scan sonar images".

为增强生物地理学优化算法(biogeography-based optimization,BBO)的优化能力并克服其不能很好平衡开发能力与避免陷入局部最优解之间的矛盾,提出基于微扰动和混合变异的差分生物地理学优化算法(differential biogeography optimization algorithm based on micro-perturbation and mixed variation,MDEBBO).引入差分变异算子和自适应的微扰动因子来改进迁移算子,使算法朝着最优解快速移动,提高算法的查找精度.采用混合变异算子代替原变异算子,在迭代前期算法具有良好的全局探索能力,在后期具有较优的局部开发性.基准测试函数的仿真结果表明了MDEBBO算法的有效性.通过MDEBBO算法对Richards模型进行参数估计预测谷氨酸菌体生长浓度,实验结果表明,MDEBBO算法较对比算法更适用于Richards模型的参数估计.

Using MODIS standard products, the temporal and spatial distribution characteristics of thermal anomalies in Henan Province in the past 12 years (2008~2019) were studied. The results found that in terms of spatial distribution, thermal anomalies were mostly concentrated in Luohe, Zhumadian, Pingdingshan, Puyang, and Shangqiu. The number of areas under the jurisdiction of Anyang, Hebi, Nanyang and Xinyang is relatively high. On the inter-annual trend, the number of thermal anomalies continued to increase from 2009 to 2013, with an average annual growth rate of 28.3%, and a continuous decline from 2013 to 2018. The decline rate was 18.4%, during which the number reached a peak of 5,843 in 2013. In terms of seasonal changes, the number of summer thermal anomalies is the largest, at 25,361, and thermal anomalies of summer are mostly concentrated in most areas of Zhumadian, Pingdingshan, Puyang and Shangqiu; the number of thermal anomalies in winter is the least, 3974, which are relatively mostly distributed in the mountainous areas of Nanyang and Xinyang. This study helps to understand the forest fires in typical areas in Henan Province, as well as heat caused by straw burning, industrial emissions, etc. Thermal anomalies changes provide technical support for regional disaster prevention and environmental monitoring.

Various applications of the Internet of Things assisted by deep learning such as autonomous driving and smart furniture have gradually penetrated people’s social life. These applications not only provide people with great convenience but also promote the progress and development of society. However, how to ensure that the important personal privacy information in the big data of the Internet of Things will not be leaked when it is stored and shared on the cloud is a challenging issue. The main challenges include (1) the changes in access rights caused by the flow of manufacturers or company personnel while sharing and (2) the lack of limitation on time and frequency. We propose a data privacy protection scheme based on time and decryption frequency limitation that can be applied in the Internet of Things. Legitimate users can obtain the original data, while users without a homomorphic encryption key can perform operation training on the homomorphic ciphertext. On the one hand, this scheme does not affect the training of the neural network model, on the other hand, it improves the confidentiality of data. Besides that, this scheme introduces a secure two-party agreement to improve security while generating keys. While revoking, each attribute is specified for the validity period in advance. Once the validity period expires, the attribute will be revoked. By using storage lists and setting tokens to limit the number of user accesses, it effectively solves the problem of data leakage that may be caused by multiple accesses in a long time. The theoretical analysis demonstrates that the proposed scheme can not only ensure safety but also improve efficiency.

The research field of automated geometry theorem proving has developed many new methods; but, all of them have not used the rings of vector. In the paper, the authors have proposed a new approach based on vector rings, implemented a machine proving program, which emphasis loop of vectors. This program could construct most common constructive geometry drawings very quickly, do automated reasoning with various vector methods according to different types of constructions which includes equal vectors, perpendicular vectors or definite proportional division points, and the proofs are concise and readable. The prover with vectors has been used to produce short and elegant proofs for some constructive constructions. Therefore, this new approach could be used in education. With many instances test, it shows automated reasoning with vectors is available, which also enhance the efficiency and readability.

In this paper, we firstly present a block robust structured multifrontal factorization method (in brief, BRSMF) using block diagonalonal structure of three temperature matrices, and then we propose a multi-core parallelization of BRSMF (in brief, MBRSMF) method based on the current mainstream parallel computer multi-core architecture. MBRSMF method parallelizes the nested dissection ordering, symbolic decomposition and numerical decomposition of BRSMF method, which aims to effectively solve three temperature linear systems on the multi-core computer. The multi-core parallelization of symbolic decomposition and numerical decomposition is based on the binary elimination tree. Theoretical analysis proves MBRSMF method has better load balancing capability. Numerical experiments show that the MBRSMF method is effective.

A novel image encryption scheme based on deoxyribonucleic acid (DNA) is proposed utilizing the hash function and coupled map lattices (CML) based on the piecewise linear chaotic map (PWLCM) in this paper. First, the chaotic sequences for the entire encryption process are generated by the PWLCM map-based CML chaotic system, and the external keys and hash value of the plain image are employed to calculate the control parameters and initial values of the CML system and PWLCM map. Especially the f(x) sequences generated by PWLCM map are used many times. Second, in accordance with the chaotic sequences produced by CML and PWLCM map, the encryption process is divided into three modules. Module one is to implement pixel-level encryption through sort function and exclusive OR (XOR) operation. Then, the DNA encoding and decoding rules are dynamic selected by chaotic sequences. DNA-level encryption is carried out in module two by cyclic shift function and dynamic DNA permutation rules. Finally, a second diffusion encryption at pixel level is performed in module three through XOR operation to further enhance the utilization of chaotic sequences and security of the image encryption system. The results of experiment and security analyses have certified that the proposed scheme has an outstanding property and can withstand a variety of typical attacks.

Role-based access control (RBAC) can effectively guarantee the security of user system data. With its good flexibility and security, RBAC occupies a mainstream position in the field of access control. However, the complexity and time-consuming of the role establishment process seriously hinder the development and application of the RBAC model. The introduction of the assistant interactive question answering algorithm based on attribute exploration (semiautomatic heuristic way to build an RBAC system) greatly reduces the complexity of building a role system. However, there are some defects in the auxiliary interactive Q&A algorithm based on attribute exploration. The algorithm is not only unable to support multiperson collaborative work but also difficult to find qualified Q&A experts in practical work. Aiming at the above problems, this paper proposes a model collaborative learning and exploration of RBAC roles under the framework of attribute exploration. In this model, after interactive Q&A with experts in different permissions systems by using attribute exploration, the obtained results are merged and calculated to get the correct role system. This model not only avoids the time-consuming process of role requirement analysis but also provides a feasible scheme for collaborative role discovery in multidepartment permissions.