高光谱图像分类的ReliefF-RFE特征选择算法构建与应用

高光谱图像具有波段连续、维数高、数据量大、相邻波段相关性强的特点,可为地物分类提供更为丰富的细节信息。但是,数据中存在大量冗余信息与噪声,在图像分类中如直接利用其所有波段特征而不进行有效分析与选择,将会导致较低的计算效率和较高的计算复杂度,分类精度亦可能随着波段维数增加而出现先增后减的“休斯（Hughes）现象”。为快速地从高达数十个甚至数百个波段的高光谱图像中提取出具有较好识别能力的特征子集,从而避免“维度灾难”,将过滤式ReliefF算法和封装式特征递归消除算法（RFE）相结合,构建了ReliefF-RFE特征选择算法,可用于高光谱图像分类的特征选择。该算法根据权重阈值,利用ReliefF算法快速剔除大量无关特征,缩小并优化特征子集的范围;利用RFE算法进一步搜索最优特征子集,将缩小范围后的特征子集中与分类器关联性小、冗余的特征进行递归筛选,进而得到分类性能最佳的特征子集。采用Indian pines数据集、Salinas-A数据集与KSC数据集等3个标准数据集作为实验数据,将ReliefF-RFE算法的应用效果与ReliefF和RFE算法进行对比。结果显示,在3个数据集中,应用ReliefF-RFE算法的高光谱图像分类平均总体精度（OA）为92.94%、F-measure为92.81%,Kappa系数为91.94%;ReliefF-RFE算法的平均特征维数是ReliefF算法的37%,而平均运算时间则是RFE算法的75%。由此表明,ReliefF-RFE算法能够在保证分类精度的同时,克服过滤式ReliefF算法无法有效减小特征之间冗余以及封装式RFE算法时间复杂度较高的缺陷,具有更为均衡的综合性能,适用于高光谱图像分类的特征选择。     

基于标准序列浮动前向特征选择的改进算法研究

随着信息技术的高速发展,信息特征的表述方法和内涵不断扩充,高维特征大幅涌现。这些高维特征中可能存在许多不相关和冗余特征,造成了维度灾难,对分类识别算法提出了更高的要求,需要利用特征选择算法,降低特征向量维数并消除数据噪音的干扰。针对高维特征向量引入的维度灾难等问题,围绕目标分类识别的具体应用,对标准的序列浮动前向特征选择算法进行了研究,并通过优化正确分类样本数目的置信上限及交叉验证的重复次数,提出了一种改进的序列浮动前向特征选择算法。通过仿真实验表明,在利用贝叶斯分类器开展识别时,改进算法能够在确保分类识别正确率的前提下,有效提升特征选择的计算速度,并随着特征选择步骤的增加,能够维持一个相对更为收敛且稳定的置信区间,具备良好的准确度。     

基于支持向量机(SVM)特征加权/选择的光谱匹配算法

高光谱数据波段多、冗余大,为了提高数据的分析效率和精度,降维是一个关键步骤。文章在文献(参考了后面的文献[18])研究的基础上,引入了迭代SVM特征选择/加权算法,为多目标遗传优化获取最优参考光谱提供一个包含有效分类信息的低维空间。基于Indiana-AVIRIS高光谱数据的实验表明,特征加权/选择的引入使光谱匹配分类精度提高了13%(相对于无特征选择的情况而言)。文章还根据光谱样本距SVM分类面的远近,定义和计算了局部权重,不仅细致刻画了同类光谱样本在局部特征空间中的分布,还使光谱相似度的计算更加灵活化,精度提高幅度达到了17%(相对于无特征选择的情况而言)。文章研究方法的提出推进了SVM在光谱数据分析中的应用深度和广度。     

新型高光谱图像快速实时目标检测与分类方法

针对逐像元处理的高光谱图像实时线性约束最小方差(LCMV)检测与分类算法计算量大、运行速度慢的问题,在LCMV检测与分类算法的基础上,提出了两种逐行的实时LCMV目标检测与分类算法。首先对LCMV算法进行了因果化,提出了逐行处理的实时因果LCMV(CR-LCMV)检测与分类算法,再利用Woodbury引理,推导出了逐行处理的实时递归因果LCMV(RCR-LCMV)检测与分类算法。实验结果表明:与LCMV检测与分类算法相比,两种新型实时算法均能在不影响检测精度的情况下实时地检测目标与对目标进行分类,且所需的数据存储空间大大降低;与逐像元处理的实时LCMV算法相比,两种新型实时算法可获得几乎与之相同的检测精度,计算复杂度大大降低,实时处理能力更强,算法在运行时间上具有明显的优越性。     

基于区域增长的图像跟踪算法的研究

为了提高序列图像的跟踪精度,提出了基于区域增长的特征量提取方法。这种方法可捕获图像中所有单连通域,并能准确地提取其特征量,然后依据所提取的特征量设计分类器,实现对图像中所有单连通域进行分类识别并加以跟踪的目的。此算法有效地解决了一般传统识别算法难以区分目标和其近邻区域的干扰,而导致跟踪目标特征量提取不准确的问题。在目标特征提取识别算法的基础上,还提出了阈值预测分割算法,并应用在序列图像的跟踪中,取得了较好效果。     

结合机器学习算法提高从头算方法对HF/HBr/H35Cl/Na35Cl振动能谱的预测性能

高精度振动能谱蕴含着分子体系的大量量子特征,是人们认识和操控分子的重要基础数据.目前,从头算方法在计算分子的振动能谱方面取得了大量成果,但是仍然面临着精度和计算量上的挑战.本文提出了一种综合从头算方法与机器学习算法进行能谱预测的新方法,在提高振动能级精度的同时大幅降低了计算量.针对HF、HBr、H³⁵Cl和Na³⁵Cl等卤素分子的研究结果表明,相较于单独的CCSD(T)/cc-pV5Z计算方法,新方法将误差减少了50%以上,同时将计算量降低了一个数量级.     

空-谱二维蚁群组合优化SVM的高光谱图像分类

提出了一种空-谱二维特征蚁群组合优化支持向量机的高光谱图像分类算法。利用两类蚁群分别在光谱维空间和样本分布空间交替搜索最大类间距波段组合和异质样本,提取最优特征波段,降低了高光谱的波段信息冗余,去除训练样本中的异质样本,优化了训练样本特征空间分布。将蚁群组合优化后的高光谱图像和训练样本应用到支持向量机(SVM)分类器中,扩大了特征空间类间距,提高了SVM算法的分类精度。实验表明该算法总分类精度达95.45%,Kappa系数0.925 2,是一种分类精度较高的高光谱图像分类方法。     

基于局部Gabor二进制模式的高光谱伪装目标图像分类

为提高光谱伪装目标图像分类精度,提出了一种基于局部Gabor二进制模式（LGBP)的空间分类方法。LGBP作为一种多尺度算法,被用来提取高光谱图像的纹理特征。然后高光谱图像中的每一个像元可以用一个光谱特征向量及一个纹理特征向量表示。通过这种方法,增大类间距离。最后使用多核支持向量机结合光谱信息和空间纹理信息实现对高光谱伪装目标图像的分类。实验证明了该方法的有效性,分类总体精度和Kappa系数分别达到了95.6%和0.937。所提出的方法对于提高分类精度及鲁棒性具有重要意义。     

基于空谱字典的加权联合稀疏表示高光谱图像分类

稀疏表示广泛用于高光谱图像分类任务中。针对字典原子空间信息和光谱信息未得到充分利用的问题,提出了基于空谱字典的加权联合稀疏表示高光谱图像分类算法。计算测试像元与字典原子的空谱联合距离,选择相似度最高的K个字典原子,并将被选择字典原子的超像素邻域扩充到新的字典中,形成空谱字典。在联合稀疏模型中,对测试像元的超像素邻域像元使用不同的权重,在空谱字典上构建加权稀疏表示模型。基于所选的两个高光谱数据集的实验证明所提算法能够有效地提高分类精度。     

ISG-FHEV等效燃油消耗最小控制策略

针对室内复杂环境下火灾识别准确率会降低的问题,提出了一种改进的粒子群算法优化支持向量机参数进行火灾火焰识别的方法。首先在 颜色空间进行火焰图像分割,对获得的火焰图像进行预处理并提取相关特征量;其次采用PSO算法搜索SVM的最优核参数和惩罚因子,并在PSO算法中加入变异操作和非线性动态调整惯性权值的方法,加快了搜索SVM最优参数的精度和速度;然后将提取的火焰各个特征量作为训练样本输入SVM模型进行训练,并建立参数优化后的SVM分类器模型;最后将待测试样本输入SVM模型进行分类识别。算法的火灾识别准确率达到94.09%,分类效果明显优于其他分类算法。仿真结果表明,改进的PSO优化SVM算法提高了火焰识别的准确率和实时性,算法的自适应性更强,误判率更低。     

First Digits’ Shannon Entropy

Related to the letters of an alphabet, entropy means the average number of binary digits required for the transmission of one character. Checking tables of statistical data, one finds that, in the first position of the numbers, the digits 1 to 9 occur with different frequencies. Correspondingly, from these probabilities, a value for the Shannon entropy H can be determined as well. Although in many cases, the Newcomb–Benford Law applies, distributions have been found where the 1 in the first position occurs up to more than 40 times as frequently as the 9. In this case, the probability of the occurrence of a particular first digit can be derived from a power function with a negative exponent p > 1. While the entropy of the first digits following an NB distribution amounts to H = 2.88, for other data distributions (diameters of craters on Venus or the weight of fragments of crushed minerals), entropy values of 2.76 and 2.04 bits per digit have been found.     

Chalcogenide Fibers: An Overview of Selected Applications

Chalcogenide glass fibers have many unique properties that make them attractive for applications that are not possible with silica fibers. Selected applications of chalcogenide fibers in research, commerce, and the military are reviewed.     

Some mathematical aspects of Anderson localization: boundary effect,multimodality, and bifurcation

Anderson localization is a famous wave phenomenon that describes the absence of diffusion of waves in a disordered medium. Here we generalize the landscape theory of Anderson localization to general elliptic operators and complex boundary conditions using a probabilistic approach, and further investigate some mathematical aspects of Anderson localization that are rarely discussed before. First, we observe that under the Neumann boundary condition, the low energy quantum states are localized on the boundary of the domain with high probability. We provide a detailed explanation of this phenomenon using the concept of extended subregions and obtain an analytical expression of this probability in the one-dimensional case. Second, we find that the quantum states may be localized in multiple different subregions with high probability in the one-dimensional case and we derive an explicit expression of this probability for various boundary conditions. Finally, we examine a bifurcation phenomenon of the localization subregion as the strength of disorder varies. The critical threshold of bifurcation is analytically computed based on a toy model and the dependence of the critical threshold on model parameters is analyzed.     

辣椒碱与β-环糊精包合物的核磁共振研究

利用核磁共振方法研究了辣椒碱β-环糊精包合物的化学计量比、空间结构信息及其在水溶液中的自扩散系数. 通过测定不同浓度比的辣椒碱和β-环糊精混合溶液的 ¹H NMR数据,绘制Job's曲线,辣椒碱和β-环糊精的Job's曲线均在r=0.5处出现拐点. 同时测定了该包合物的2D ROESY和DOSY谱图,ROESY谱图中NOE交叉信号出现在辣椒碱的H-1~H-8和β-环糊精的H-3′、H-5′、H-6′之间,DOSY测得β-环糊精和辣椒碱形成包合物前后的表观自扩散系数. 结果表明,辣椒碱β-环糊精包合物的主客体分子的化学计量比为1∶1,辣椒碱分子的异丙基端从β-环糊精的宽口端进入疏水腔,其中H-1~H-8部分在空腔内部,包合物的自扩散系数为2.95×10^-10m²/s.     

Darboux Transformation and N-soliton Solution for Extended Form of Modified Kadomtsev-Petviashvili Equation with Variable-Coefficient

The extended form of modified Kadomtsev-Petviashvili equation with variable-coefficient is investigated in the framework of Painlevé analysis. The Lax pairs are obtained by analysing two Painlevé branches of this equation. Starting with the Lax pair, the N-times Darboux transformation is constructed and the N-soliton solution formula is given, which contains 2n free parameters and two arbitrary functions. Furthermore, with different combinations of the parameters, several types of soliton solutions are calculated from the first order to the third order. The regularity conditions are discussed in order to avoid the singularity of the solutions. Moreover, we construct the generalized Darboux transformation matrix by considering a special limiting process and find a rational-type solution for this equation.     

Thermodynamics of 4D dS/AdS Gauss–Bonnet black holes according to consistent gravity theory in the presence of a cloud of strings

By looking at the Lovelock theorem one can infer that the gravity model given by [1] cannot be applicable for all types of 4D Einstein–Gauss–Bonnet (EGB) curved space-time. The reason for this is that in 4D space-time, the Gauss–Bonnet invariant is a total derivative and hence it does not contribute to gravitational dynamics. Hence, the authors of [2] presented an alternative consistent EGB gravity model instead of [1] by applying a break-of-diffeomorphism property. In this work, we use the alternative model to produce a de Sitter (dS)/Anti-de Sitter (AdS) black hole metric and then investigate its thermodynamic behavior in the presence of a cloud of Nambu–Goto strings. Mathematical derivations show that the resulting diagrams of pressure vs specific volume at a constant temperature are similar to that for a van der Waals gas/fluid in an ordinary thermodynamic system in the dS sector but not in the AdS background. From this, we infer that the black hole participates in the small-to-large black hole phase transition in the dS background, while it exhibits a Hawking–Page phase transition in the AdS background. In the latter case, an evaporating black hole eventually reaches an AdS vacuum space because of its instability.     

Bidirectional Quantum Secure Direct Communication Network Protocol with Hyperentanglement

We propose a bidirectional quantum secure direct communication (QSDC) network protocol with the hyperentanglment in both the spatial-mode ad the polarization degrees of freedom of photon pairs which can in principle be produced with a beta barium borate crystal. The secret message can be encoded on the photon pairs with unitary operations in these two degrees of freedom independently. Compared with other QSDC network protocols, our QSDC network protocol has a higher capacity as each photon pair can carry 4 bits of information. Also, we discuss the security of our QSDC network protocol and its feasibility with current techniques.     

Chess AI: Competing Paradigms for Machine Intelligence

Endgame studies have long served as a tool for testing human creativity and intelligence. We find that they can serve as a tool for testing machine ability as well. Two of the leading chess engines, Stockfish and Leela Chess Zero (LCZero), employ significantly different methods during play. We use Plaskett’s Puzzle, a famous endgame study from the late 1970s, to compare the two engines. Our experiments show that Stockfish outperforms LCZero on the puzzle. We examine the algorithmic differences between the engines and use our observations as a basis for carefully interpreting the test results. Drawing inspiration from how humans solve chess problems, we ask whether machines can possess a form of imagination. On the theoretical side, we describe how Bellman’s equation may be applied to optimize the probability of winning. To conclude, we discuss the implications of our work on artificial intelligence (AI) and artificial general intelligence (AGI), suggesting possible avenues for future research.     

Double Diffractive Dissociation Process p + p(p) → Xp + Xp(p) + Π+ + Π- at High Energy and the Pomeron Model of Donnachie-Landshoff

Large p_γ charged pion pair produced in Double Diffractive Dissocjation (DDD) process at high energy is studied with double pomeron (IP) exchange mechanism. Using DonnachieLandshoff pomeron model we have calculated the cross-section of the process at the energy interval from ISR to LHC when the rapidity of the charged pions is approximately equal to zero. It is shown that this process is the favorite in experiments which could effectively be used to examine Donnachie-Landshoff pomeron model and provide a possibility to check the value of the suppression coefficient N².     

HL-2M 装置时序控制与数据采集系统设计

基于 FPGA 和 IEEE1588PTPv2 协议设计了 HL-2M 装置时序控制系统,用于在精准时刻为测控系统 提供触发。数据采集系统采用客户端/服务端双模式管理。在 HL-2M 装置初始等离子体放电实验中测试结果表明, 时序控制和采集系统具有纳秒级精度触发,实时性强,数据传输速度快,稳定可靠,方便部署和管理的特点,满 足 HL-2M 等离子体放电实验的应用需求。