A Spectral Convolutional Neural Network Model Based on Adaptive Fick’s Law for Hyperspectral Image Classification

Hyperspectral image classification stands as a pivotal task within the field of remote sensing,yet achieving highprecision classification remains a significant challenge.In response to this challenge,a Spectral Convolutional Neural Network model based on Adaptive Fick’s Law Algorithm(AFLA-SCNN)is proposed.The Adaptive Fick’s Law Algorithm(AFLA)constitutes a novel metaheuristic algorithm introduced herein,encompassing three new strategies:Adaptive weight factor,Gaussian mutation,and probability update policy.With adaptive weight factor,the algorithmcan adjust theweights according to the change in the number of iterations to improve the performance of the algorithm.Gaussianmutation helps the algorithm avoid falling into local optimal solutions and improves the searchability of the algorithm.The probability update strategy helps to improve the exploitability and adaptability of the algorithm.Within the AFLA-SCNN model,AFLA is employed to optimize two hyperparameters in the SCNN model,namely,“numEpochs”and“miniBatchSize”,to attain their optimal values.AFLA’s performance is initially validated across 28 functions in 10D,30D,and 50D for CEC2013 and 29 functions in 10D,30D,and 50D for CEC2017.Experimental results indicate AFLA’s marked performance superiority over nine other prominent optimization algorithms.Subsequently,the AFLA-SCNN model was compared with the Spectral Convolutional Neural Network model based on Fick’s Law Algorithm(FLA-SCNN),Spectral Convolutional Neural Network model based on Harris Hawks Optimization(HHO-SCNN),Spectral Convolutional Neural Network model based onDifferential Evolution(DE-SCNN),SpectralConvolutionalNeuralNetwork(SCNN)model,and SupportVector Machines(SVM)model using the Indian Pines dataset and PaviaUniversity dataset.The experimental results show that the AFLA-SCNN model outperforms other models in terms of Accuracy,Precision,Recall,and F1-score on Indian Pines and Pavia University.Among them,the Accuracy of the AFLA-SCNN model on Indian Pines reached 99.875%,and the Accuracy on PaviaUniversity reached 98.022%.In conclusion,our proposed AFLA-SCNN model is deemed to significantly enhance the precision of hyperspectral image classification.

Targeted isolation and identification of bioactive pyrrolidine alkaloids from Codonopsis pilosula using characteristic fragmentation-assisted mass spectral networking

Codonopsis pilosula(CP),a well-known food medicine homology plant,is commonly used in many countries.In our preliminary study,a series of pyrrolidine alkaloids with high MS responses were detected as characteristic absorbed constituents in rat plasma after oral administration of CP extract.However,their structures were unclear due to the presence of various isomers and the lack of reference standards.In the present study,an MS-guided targeted isolation of pyrrolidine alkaloids of CP extract was performed by ultra-performance liquid chromatography coupled with quadrupole time-of-flight tandem mass spectrometry(UPLC/Q-TOF MS).For data analysis under fast data directed acquisition mode(Fast-DDA),an effective approach named characteristic fragmentation-assisted mass spectral networking was successfully applied to discover new pyrrolidine alkaloids with high MS response in CP extract.As a result,seven new pyrrolizidine alkaloids[codonopyrrolidiums C–I(3–9)],together with two known ones(1 and 2),were isolated and identified by NMR spectral analysis.Among them,codonopyrrolidium B(1),codonopyrrolidium D(4)and codonopyrrolidium E(5)were evaluated for lipid-lowering activity,and they could improve high fructose-induced lipid accumulation in HepG2 cells.In addition,the characteristic MS/MS fragmentation patterns of these pyrrolizidine alkaloids were investigated,and 17 pyrrolidine alkaloids were identified.This approach could accelerate novel natural products discovery and characterize a class of natural products with MS/MS fragmentation patterns from similar chemical scaffolds.The research also provides a chemical basis for revealing in vivo effective substances in CP.

Deep Spectrum Prediction in High Frequency Communication Based on Temporal-Spectral Residual Network

High frequency(HF) communication is widely spread due to some merits like easy deployment and wide communication coverage. Spectrum prediction is a promising technique to facilitate the working frequency selection and enhance the function of automatic link establishment. Most of the existing spectrum prediction algorithms focus on predicting spectrum values in a slot-by-slot manner and therefore are lack of timeliness. Deep learning based spectrum prediction is developed in this paper by simultaneously predicting multi-slot ahead states of multiple spectrum points within a period of time. Specifically, we first employ supervised learning and construct samples depending on longterm and short-term HF spectrum data. Then, advanced residual units are introduced to build multiple residual network modules to respectively capture characteristics in these data with diverse time scales. Further, convolution neural network fuses the outputs of residual network modules above for temporal-spectral prediction, which is combined with residual network modules to construct the deep temporal-spectral residual network. Experiments have demonstrated that the approach proposed in this paper has a significant advantage over the benchmark schemes.

Scaling of weighted spectral distribution in weighted small-world networks

Many real-world systems can be modeled by weighted small-world networks with high clustering coefficients. Recent studies for rigorously analyzing the weighted spectral distribution（W SD） have focused on unweighted networks with low clustering coefficients. In this paper, we rigorously analyze the W SD in a deterministic weighted scale-free small-world network model and find that the W SD grows sublinearly with increasing network order（i.e., the number of nodes） and provides a sensitive discrimination for each input of this model. This study demonstrates that the scaling feature of the W SD exists in the weighted network model which has high and order-independent clustering coefficients and reasonable power-law exponents.

Hierarchical Modeling by Recursive Unsupervised Spectral Clustering and Network Extended Importance Measures to Analyze the Reliability Characteristics of Complex Network Systems

The complexity of large-scale network systems made of a large number of nonlinearly interconnected components is a restrictive facet for their modeling and analysis. In this paper, we propose a framework of hierarchical modeling of a complex network system, based on a recursive unsupervised spectral clustering method. The hierarchical model serves the purpose of facilitating the management of complexity in the analysis of real-world critical infrastructures. We exemplify this by referring to the reliability analysis of the 380 kV Italian Power Transmission Network (IPTN). In this work of analysis, the classical component Importance Measures (IMs) of reliability theory have been extended to render them compatible and applicable to a complex distributed network system. By utilizing these extended IMs, the reliability properties of the IPTN system can be evaluated in the framework of the hierarchical system model, with the aim of providing risk managers with information on the risk/safety significance of system structures and components.

JOINT RESOURCE ALLOCATION FOR WLAN&WCDMA INTEGRATED NETWORKS BASED ON SPECTRAL BANDWIDTH MAPPING

Next wireless network aims to integrate heterogeneous wireless access networks by sharing wireless resource.The spectral bandwidth mapping concept is proposed to uniformly describe the resource in heterogeneous wireless networks.The resources of codes and power levels in WCDMA system as well as statistical time slots in WLAN are mapped into equivalent bandwidth which can be allocated in different networks and layers.The equivalent bandwidth is jointly distributed in call admission and vertical handoff control process in an integrated WLAN/WCDMA system to optimize the network utility and guarantee the heterogeneous QoS required by calls.Numerical results show that,when the incoming traffic is moderate,the proposed scheme could receive 5%-10% increase of system revenue compared to the MDP based algorithms.

Clustering in Wireless Multimedia Sensor Networks Using Spectral Graph Partitioning

Wireless multimedia sensor network (WMSN) consists of sensors that can monitor multimedia data from its surrounding, such as capturing image, video and audio. To transmit multimedia information, large energy is required which decreases the lifetime of the network. In this paper we present a clustering approach based on spectral graph partitioning (SGP) for WMSN that increases the lifetime of the network. The efficient strategies for cluster head selection and rotation are also proposed.

Influence of Topological Properties of Complex Networks on the Effect of Spectral Coarse-Grained Network

Recently, some coarse-graining methods based on network synchronization have been proposed to reduce the network size while preserving the synchronizability of the original network. In this paper, we investigate the effects of the coarse graining process on synchronizability over complex networks under different average path lengths and different degrees of distribution. A large amount of experiments demonstrate a close correlation between the average path length, the heterogeneity of the degree distribution and the ability of spectral coarse-grained scheme in preserving the network synchronizability. We find that synchronizability can be well preserved in spectral coarse-grained networks when the considered networks have a longer average path length or a larger degree of variance.

Secure Spectral-Energy Efficiency Tradeoff in RandomCognitive Relay Networks

Spectral efficiency(SE) and energy efficiency(EE) in secure communications is of primary importance due to the fact that 5 G wireless networks aim to achieve high throughput,low power consumption and high level of security.Nevertheless,maximizing SE and EE are not achievable simultaneously.In this paper,we investigate the SE and EE tradeoff for secure transmission in cognitive relay networks where all nodes are randomly distributed.We first introduce the opportunistic relay selection policy,where each primary transmitter communicates with the primary receiver with the help of a secondary user as a relay.Then,we evaluate the secure SE and secure EE of the primary network based on the outage probabilities analysis.Thirdly,by applying a unified SE-EE tradeoff metric,the secure SE and EE tradeoff problem is formulated as the joint secure SE and EE maximization problem.Considering the non-concave feature of the objective function,an iterative algorithm is proposed to improve secure SE and EE tradeoff.Numerical results show that the opportunistic relay selection policy is always superior to random relay selection policy.Furthermore,the opportunistic relay selection policy outperforms conventional direct transmission policy when faced with small security threat(i.e.,for smaller eavesdropper density).

图卷积神经网络综述

图卷积神经网络是图论与深度学习的交叉,已成为机器学习领域的研究热点。基于此,介绍了图卷积神经网络的形成,梳理了两大类经典的图卷积神经网络:谱方法和空间方法,详细介绍了这两类图卷积神经网络模型,分析了图卷积操作的核心理论基础,介绍了图卷积神经网络在各领域的应用,总结了图卷积神经网络面临的主要挑战,展望了图卷积神经网络的发展趋势,并分析了图卷积神经网络在野外环境下蝴蝶识别任务中的潜在应用。

The Monitoring of Red Tides Based on Modular Neural Networks Using Airborne Hyperspectral Remote Sensing

这篇论文建议基于聚类和模块化的神经网络监视方法的红潮。从天线的一个团获得红潮的特征遥远的 sensinghyperspectral 数据，首先，日志剩余修正(纵向冗余码校验) 被用来使数据，然后聚类的分析正常化被采用为神经网络选择并且形成训练样本。监视的 Forrapid，辨别者由模块化的神经网络组成，其结构和学习参数被一个适应基因算法(统帅) 决定。实验证明这个方法罐头很快并且有效地监视红潮。

基于拆分注意力网络的单图像超分辨率重建

针对现有生成对抗网络的单图像超分辨率重建在大尺度因子下存在训练不稳定、特征提取不足和重建结果纹理细节严重缺失的问题,提出一种拆分注意力网络的单图超分辨率重建方法。首先,以拆分注意力残差模块作为基本残差块构造生成器,提高生成器特征提取的能力。其次,在损失函数中引入鲁棒性更好的Charbonnier损失函数和Focal Frequency Loss损失函数代替均方差损失函数,同时加入正则化损失平滑训练结果,防止图像过于像素化。最后,在生成器和判别器中采用谱归一化处理,提高网络的稳定性。在4倍放大因子下,与其他方法在Set5、Set14、BSDS100、Urban100测试集上进行测试比较,本文方法的峰值信噪比比其他对比方法的平均值提升1.419 dB,结构相似性比其他对比方法的平均值提升0.051。实验数据和效果图表明,该方法主观上具有丰富的细节和更好的视觉效果,客观上具有较高的峰值信噪比值和结构相似度值。

Stability of weighted spectral distribution in a pseudo tree-like network model

The comparison of networks with different orders strongly depends on the stability analysis of graph features in evolving systems. In this paper, we rigorously investigate the stability of the weighted spectral distribution（i.e., a spectral graph feature） as the network order increases. First, we use deterministic scale-free networks generated by a pseudo treelike model to derive the precise formula of the spectral feature, and then analyze the stability of the spectral feature based on the precise formula. Except for the scale-free feature, the pseudo tree-like model exhibits the hierarchical and small-world structures of complex networks. The stability analysis is useful for the classification of networks with different orders and the similarity analysis of networks that may belong to the same evolving system.

基于机器学习的语音增强技术

主要研究基于机器学习的语音增强技术,以提升语音信号的质量。首先,介绍基于机器学习的语音增强系统框架。其次,详细探讨谱减法与深度神经网络(Deep Neural Network,DNN)相结合的语音增强方法的数学原理。最后,采用NOISEX-92数据集测试与评估提出的方法。实验结果表明,基于谱减法与DNN的语音增强方法在提升信噪比和语音清晰度方面取得显著的效果,能够有效提升语音通信质量。

可重构智能表面辅助通信系统网络架构演进综述

随着无线通信技术的发展,越来越多的无线终端接入到通信系统中,对通信网络造成很大的业务承载与传输压力,尤其是障碍物严重阻挡或强信道衰落情况下,使得接收机信号以及系统频谱效率下降现象日益严峻。为了解决该问题,近年来,可重构智能表面(Reconfigurable Intelligent Surface, RIS)作为一种有效的6G候选技术被提出。RIS可以通过电磁调控方式主动改变信道传输质量,从而能够解决无线通信系统频谱效率受限、信号补盲、绕障通信等现实难题。基于此,对RIS辅助通信网络架构演进进行了研究。介绍了RIS的基本概念,并对其三种基本网络架构进行了分析与对比;根据不同的信号传输类型和传输环境,对RIS辅助通信网络架构进行了分析与设计;对当前网络架构发展所面临的挑战以及未来研究趋势进行了展望,为RIS辅助通信系统性能分析、资源分配、网络优化、RIS位置部署提供帮助。

面向天文多普勒差分测速的太阳/行星光谱对生成方法

为了提供天文多普勒差分测速所需的同步太阳/行星光谱对,提出了一种变分自编码器(VAE)和对偶生成对抗网络(Dual GAN)相融合的VAE-Dual GAN。首先,实测太阳光谱经过VAE编码到隐空间,实现了光谱到光谱域的扩充;然后,由Dual GAN将隐空间映射到伪行星光谱;最后,利用伪行星光谱生成重构太阳光谱。此外,利用编码和生成重建损失加强对网络的约束。VAE-Dual GAN利用Dual GAN的转换学习能力完成了两个光谱域的转换,生成同步太阳/行星光谱对。实验结果表明,VAE-Dual GAN可生成高质量的太阳/行星光谱对,将天文多普勒差分测速精度提高60%以上。

复谱映射下融合高效Transformer的语音增强方法

针对卷积神经网络(Convolutional Neural Network,CNN)过去在语音增强中表现优异但对全局特征捕获不足,以及Transformer近年展现出长序列间依赖优势但又存在局部细节特征丢失、参数量大等问题,该文为了充分利用CNN与Transformer的优势并弥补各自不足,提出了一种在复频谱映射下的新型卷积模块与高效Transformer融合的单通道语音增强网络。该网络由编码层、传输层与双分支解码层组成:在编解码部分设计了一种协作学习模块(Collaborative Learning Block,CLB)来监督交互信息,在减少参数量的同时提高主干网络对复特征的获取能力;传输层中则提出一种时频空间注意Transformer模块分别对语音子频带和全频带信息建模,充分利用声学特性来模拟局部频谱模式并捕获谐波间依赖关系。将该模块进一步与通道注意分支相结合,设计了一种可学习的双分支注意融合(Dual-branch Attention Fusion,DAF)机制,从空间-通道角度提取上下文特征以加强信息的多维度传输;最后,在此基础上搭建一种高斯加权渐进网络作为中间传输层,通过堆叠DAF模块进行加权求和后输出以充分利用深层特征,使得解码过程更具鲁棒性。分别在英文VoiceBank-DEMAND数据集、中文THCHS30语料库与115种环境噪声下进行消融以及综合对比实验,结果表明,该文方法仅以最小0.68×10^(6)的参数量,相比于大部分最新相关网络模型取得了更优的主、客观指标,具有较为突出的增强性能与泛化能力。

基于谱聚类的主动配电网多时间尺度无功优化策略

高比例分布式光伏接入配电网后,传统优化方案无法有效平抑电压波动,分布式光伏逆变器的无功调控能力难以充分利用。为此,提出一种基于谱聚类的主动配电网多时间尺度无功优化策略,该方法分为日前优化和日内实时优化两个阶段。首先,对离散设备的时间耦合性进行解耦,以配电网网损、平均电压偏差、电压波动严重程度为目标函数,建立基于社交网络搜索算法的日前无功优化模型,确定离散设备静态最优档位序列;其次,通过谱聚类的方法进行耦合,确定离散设备动态最优档位序列,结合改进的分布式光伏逆变器就地控制策略,建立日内实时优化模型,从而抑制日前预测数据偏差导致的电压波动;最后,基于改进后的IEEE33节点系统进行仿真实验。仿真结果表明,所提策略可以有效降低运算难度、提高求解效率,验证了该策略的有效性和优越性。

基于CNN-NLSTM的脑电信号注意力状态分类方法

通过脑电信号进行注意力状态检测,对扩大脑-机接口技术的应用范围具有重要意义。为了提高注意力状态的分类准确率,该文提出一种基于CNN-NLSTM的脑电信号分类模型。首先采用Welch方法获得脑电信号的功率谱密度特征并将其表示为二维灰度图像。然后使用卷积神经网络从灰度图像中学习表征注意力状态的特征,并将相关特征输入到嵌套长短时记忆神经网络依次获得所有时间步骤的注意力特征。最后将两个网络依次连接来构建深度学习框架进行注意力状态分类。实验结果表明,该文所提出的模型通过进行多次5-折交叉验证评估后得到89.26%的平均分类准确率和90.40%的最大分类准确率,与其他模型相比具有更好的分类效果和稳定性。

卷积神经网络结合改进光谱处理方法用于马铃薯病害检测

针对马铃薯早疫病不同染病时期光谱数据受到杂散光和噪声等因素的干扰,以及波段数众多、数据量大且谱带复杂会对光谱的定量和定性分析产生不利影响,研究9种光谱预处理方法,结合实验结果,将预处理方法进行排列组合,扩展改进为16种光谱预处理方法,并与连续投影算法、竞争自适应重加权算法和遗传算法3种特征波段提取方法进行组合得到64种光谱处理方法对原始光谱数据进行优化处理。在卷积神经网络(CNN)分类模型中,大部分经过光谱处理方法优化后的光谱数据分类精度相比原始数据的总体分类精度86.67%明显提高,其中12种光谱处理方法的分类精度达到100%,实现对马铃薯早疫病不同染病时期的理想分类。为进一步对马铃薯早疫病不同染病时期进行定量分析,将经过光谱处理方法处理后的光谱数据使用构建的CNN定量估算模型进行定量分析,结果表明,光谱预处理在优化数据的同时,也会损失数据中对目标变量有用的光谱信息,从而导致经过光谱分析方法处理后的数据结果相对于原始光谱数据的R^(2)和RMSE会出现下降的结果,通过研究使用的融合光谱处理方法对原始光谱数据优化能够进一步提升模型性能,其中基于均值中心化、多元散射校正、移动平均平滑相结合的光谱处理方法的CNN定量估算模型取得了最好的结果,其R^(2)为1说明估算的马铃薯早疫病不同染病时期和实际值拟合程度达到100%拟合,其RMSE仅为0.001 1,表明马铃薯早疫病不同染病时期的估算值与真实值之间的偏差接近0,说明该模型能够对马铃薯早疫病不同染病时期进行完美预测。结果表明提出的CNN能够对马铃薯早疫病不同染病时期进行有效地分类检测和定量分析,将各类预处理和特征波段提取方法按优化目的进行有效组合能够有效提高建模效果,为农作物病害无损、精准、智能化检测提供理论和技术支持。