A novel kernel Fisher discriminant analysis: constructing informative kernel by decision tree ensemble for metabolomics data analysis

Large amounts of data from high-throughput metabolomics experiments become commonly more and more complex, which brings an enormous amount of challenges to existing statistical modeling. Thus there is a need to develop statistically efficient approach for mining the underlying metabolite information contained by metabolomics data under investigation. In the work, we developed a novel kernel Fisher discriminant analysis (KFDA) algorithm by constructing an informative kernel based on decision tree ensemble. The constructed kernel can effectively encode the similarities of metabolomics samples between informative metabolites/biomarkers in specific parts of the measurement space. Simultaneously, informative metabolites or potential biomarkers can be successfully discovered by variable importance ranking in the process of building kernel. Moreover, KFDA can also deal with nonlinear relationship in the metabolomics data by such a kernel to some extent. Finally, two real metabolomics datasets together with a simulated data were used to demonstrate the performance of the proposed approach through the comparison of different approaches.     

Kernels by monochromatic paths in digraphs with covering number 2

We call the digraph D an k-colored digraph if the arcs of D are colored with k colors. A subdigraph H of D is called monochromatic if all of its arcs are colored alike. A set N⊆V(D) is said to be a kernel by monochromatic paths if it satisfies the following two conditions: (i) for every pair of different vertices u,v∈N, there is no monochromatic directed path between them, and (ii) for every vertex x∈(V(D)?N), there is a vertex y∈N such that there is an xy-monochromatic directed path. In this paper, we prove that if D is an k-colored digraph that can be partitioned into two vertex-disjoint transitive tournaments such that every directed cycle of length 3,4 or 5 is monochromatic, then D has a kernel by monochromatic paths. This result gives a positive answer (for this family of digraphs) of the following question, which has motivated many results in monochromatic kernel theory: Is there a natural numberlsuch that if a digraphDisk-colored so that every directed cycle of length at mostlis monochromatic, thenDhas a kernel by monochromatic paths?     

高光谱的刺五加黑斑病的早期检测研究

对感染黑斑病的刺五加叶片进行光谱特性研究,能为药用植物病害的早期筛选与精准治疗提供重要研究资料。实验目的,运用高光谱成像技术实现植物病害的自动监督分类与识别。实验过程,首先使用高光谱成像系统在可见光波段（380～960 nm）内采集刺五加黑斑病的叶片样本,光谱数据经过去除亮暗噪声和平滑预处理后,再经过主成分分析实现数据降维,继而运用基于不同核函数的支持向量机法建立分类模型,最后利用总体分类精度、Kappa系数等因子评价不同核函数对分类器性能的影响。根据叶片表面的特征将其分为四类样本：健康亮部、健康暗部、轻度病害和重度病害等。对比各类样本的光谱可知,刺五加的健康样本在540 nm波长存在一个明显峰值,在620～680 nm光谱曲线急剧上升;而病害样本的光谱反射率呈现缓慢且平稳的上升趋势,上述特征能够将图像空间上反射强度接近的健康亮部和严重病害完全区分开。经对比发现前四个主成分（PC1,PC2,PC3,PC4）在分类表达上存在差异,主要表现为PC1含有的信息多,能够较好地区分各类样本;PC2则出现健康亮部和严重病害的交叉混淆;PC3是对于PC2的补充,能基本完整地表达轻微病害;PC4的贡献率仅有0.19%,依然能够准确地识别严重病害。不同主成分分量在表达各类样本特征中存在的差异能够作为复杂样本分类的参考依据。对比四种核函数对支持向量机分类器性能的影响,结果显示线性核函数的识别过程受光强反射的影响较大,Sigmoid核函数的训练精度易受数据集大小的影响,在识别健康亮或暗,以及轻微病害上均存在一定的误差,多项式核函数与径向基核函数的效果较好,其中,多项式核函数的精度更高,为92.77%。研究表明,利用高光谱成像技术能够准确地识别刺五加的健康叶片和患病叶片,为实现自动诊断药用植物叶片病害提供新方法。     

一种基于Bhattacharyya核的SVM

包括图像识别在内的很多应用领域里,把单个样本表示成向量的集合的形式是很自然的想法,利用一个合适的核函数我们可以把这些向量映射到一个更高维的Hilbert空间,在这个高维空间里用Kernel PCA方法找到样本的高斯分布族,这样就可以把样本上的核函数定义成它们所服从的高斯分布密度函数的Bhattacharrya仿射.这样得到的核函数具有比较好的性质,比如说在各种变换下有稳定性表现,从而也说明了即使还有别的表示样本的方法,用向量集合的形式来表示单个的样本也是具有合理性的.     

核密度中含有参数的Cauchy主值积分的求积公式

在本文中，我们以二元三次复插值样条函数作为逼近工具，给出了光滑封闭曲线上的核密度中含有参数的各种Ｃａｕｃｈｙ主值积分的求积公式，误差估计及逼近性定理。     

基于SVM理论的一种新的数据分类方法

基于 SVM分类器在模式识别问题中有独特的优势 ,本文通过对标准 SVM模型的改造 ,提出了一种新的简单的数据分类方法 .理论分析和实验表明 ,该方法与标准 SVM分类方法相比具有处理大规模数据识别的能力且保持较高的样本识别率 ,节省存储空间等优势 .     

Kernel energy method: Basis functions and quantum methods

The kernel energy method (KEM) has been illustrated with peptides and has been shown to reduce the computational difficulty associated with obtaining ab initio quality quantum chemistry results for large biological compounds. In a recent paper, the method was illustrated by application to 15 different peptides, ranging in size from 4 to 19 amino acid residues, and was found to deliver accurate Hartree–Fock (HF) molecular energies within the model, using Slater‐type orbital (STO)‐3G basis functions. A question arises concerning whether the results obtained from the use of KEM are wholly dependent on the STO‐3G basis functions that were employed, because of their relative simplicity, in the first applications. In the present work, it is shown that the accuracy of KEM does not depend on a particular choice of basis functions. This is done by calculating the ground‐state energy of a representative peptide, ADPGV7B, containing seven amino acid residues, using seven different commonly employed basis function sets, ranging in size from small to medium to large. It is shown that the accuracy of the KEM does not vary in any systematic way with the size or mathematical completeness of the basis set used, and good accuracy is maintained over the entire variety of basis sets that have been tested. Both approximate HF and density functional theory (DFT) calculations are made. We conclude that the accuracy inherent in the KEM is not dependent on a particular choice of basis functions. The first application, to 15 different peptides mentioned above, employed only HF calculations. A second question that arises is whether the results obtained with the use of KEM will be accurate only within the HF approximation. Therefore, in the present work we also study whether KEM is applicable across a variety of quantum computational methods, characterized by differing levels of accuracy. The peptide, Zaib4, containing 74 atoms, was used to calculate its energy at seven different levels of accuracy. These include the semi‐empirical methods, AM1 and PM5, a DFT B3LYP model, and ab initio HF, MP2, CID, and CCSD calculations. KEM was found to be widely applicable across the spectrum of quantum methods tested. © 2005 Wiley Periodicals, Inc. Int J Quantum Chem, 2006     

On the core and nucleolus of minimum cost spanning tree games

We develop two efficient procedures for generating cost allocation vectors in the core of a minimum cost spanning tree (m.c.s.t.) game. The first procedure requires O(n ²) elementary operations to obtain each additional point in the core, wheren is the number of users. The efficiency of the second procedure, which is a natural strengthening of the first procedure, stems from the special structure of minimum excess coalitions in the core of an m.c.s.t. game. This special structure is later used (i) to ease the computational difficulty in computing the nucleolus of an m.c.s.t. game, and (ii) to provide a geometric characterization for the nucleolus of an m.c.s.t. game. This geometric characterization implies that in an m.c.s.t. game the nucleolus is the unique point in the intersection of the core and the kernel. We further develop an efficient procedure for generating fair cost allocations which, in some instances, coincide with the nucleolus. Finally, we show that by employing Sterns' transfer scheme we can generate a sequence of cost vectors which converges to the nucleolus. Part of this research was done while the author was visiting the Department of Operations Research at Stanford University. This research was partially supported by Natural Sciences and Engineering Research Council Canada Grant A-4181.     

带核集分划问题的一个线性1／7近似算法

设有整数集Ｓ＝｛ｒ１，ｒ２；ｐ１，ｐ２，…，ｐｎ｝，这里ｒｉ≥０，ｐｊ＞０（ｉ＝１，２；ｊ＝１，２，…，ｎ），寻找一个Ｓ的最优分划Ｐ＝（Ｓ＊１，Ｓ＊２）使得：（１）ｒｉ属于不同子集，（２）Ｓ＊１与Ｓ＊２中元素总和较大者尽可能地小．这是一个ＮＰ－完备问题，本文给出一个线性时间近似算法，它的近似界为８７．     

Group decisions I. In arbitrary spaces of fuzzy binary relations

This article solves the problem of finding a set of group decisions that satisfy the classical Pareto unanimity principle for the case of initial data represented as fuzzy relations of individual preference. The solution proceeds from results obtained in studying the structure of convex (in the sense defined here) sets and their convex hulls. In the first part that study is carried out for spaces of arbitrary fuzzy binary relations.