K-近邻预测后验风险的指数限

设(X,θ),(X_i,θ_i),i=1,…,n,为iid随机向量,取值于R~d×R~1.要根据X,借助于样本Z~n={(X_i,θ_i),i=1,…,n},去预测θ之值,Cover在[1]中把k近邻方法用于这个问题。其法如下:在R~d上引进距离‖x-y‖,把X_1,…,X_n按其与X的距离重排为(当等号出现时,可采用“足标小的排前”的原则),取定自然数k,     

近邻预测风险的收■性

§1 引言和结果设(X,θ)是一个取值于 R~d×R~l 的随机向量,对其分布一无所知。(X_1,θ_1),…,(X_n,θ_n)为(X,θ)的观察样本。假定(X,θ)(X_1,θ_t),…,(X_n,θ_n)是独立同分布(iid.)的。设已有了 X 的观察值 x,但θ之值未观察。要依据样本(X_i,θ_i),i=1,…,n,及 X 的已知值 x,去预测θ的值。由于对(X,θ)的分布无所知,这个问题是非参数性的,通常的线性回归方法等都不适用。有一种简单而比较实用的非参数方法,叫近邻预测法,其法如下:先按与 x 的距离     

再论最近邻判别分析

一、引言: 设(X,θ)为在R~d×{1,…,M}中取值的随机向量,称X为指标变量,θ为类别变量。假定X的值已知,据此要对与其匹配的θ值进行判定,这就是判别分析问题。在(X,θ)的分布未知的情况下,为进行判别,除利用X的当前观察值外,还要借助于历史知识。设(X_i,θ_i),i=1,…,n,是(X,θ)的一组iid样本,常称为训练样本。此时我们考虑的问题是,利用(X_i,θ_i),i=1,…,n及X值对θ进行判别。假定在R~d中引进了某一距离函数p(x,     

多参数同时估计的容许性

令 X_1,…,X_n 是一串独立随机变量,且 X_1～P_(θ_i)θ_i∈(?)_i,(i=1,2,…,n),假设估计θ_i 的损失函数为 L(θ_i,d_i),δ_i(X_i)是仅依赖 X_i,θ_i 的一个容许估计(i=1,2,…,n).现在我们要同时估计(θ_1,…,θ_n)′(?)θ,其损失函数取为 sum from i=1 to n L(θ_i,d_i),那么(δ_i(X_1),…,δ_n(X_n))′是θ的容许估计吗?早在50年代,Stein 就证明了,在 n≥3,X_i～N(θ_i,1),L(θ_i,d_i)=(θ_i-d_i)~2条件下,上述结论不成立.近20余年,很多作者也研究了这个问题,指出 Stein 的现象对许多分布,例如 Poisson 分布,Gama 分布,负二项分布及位置参数估计皆存在.但在什么条件下,(δ,(X_1),…,δ_n(X))′是容许的则很少研究,仅仅有少数特殊情况下的结果(见[3]).本文给出了相当一般的充分条件(定理1.1),利用定理1.1,研究了 L(θ_i,d_i)=λ(θ_i)(g(θ_i)-d_i)~2时,结论成立的充分条件(定理2.1).还给出了多个位置参数,Pitman 估计为容许的充分条件.最后一节给出了五个具体例子,它包括在平方损失下,多个正态密度及分布函数的容许估计;参数自然区间 为有限区间之指数族分布,在平方损失下,同时估计多个均值的线性容许估计;若 X_i～Poisson 分布 P_(2_i),i=1,2,…,n(a_1x_1,…,a_nx_n)′在损失函数sum from i=1 to n     

随机化最近邻判别分析中错误概率估计的极限性质

设(X,θ)是取值于 R~d×{1,…,M)的随机向量.我们分别称 X 与θ为指标变量和类别变量.又设 Z~n(?){(X_i,θ_i),i=1,…,n}为(X,θ)的 iid.样本,称之为训练样本.判别分析的问题就是要依据 Z~n 及 X 的观察值对θ进行判别.假定在 R~d 中引进了某一距离ρ(x,y),x,y∈R~d.于是当 X=x 给定时,我们可按照距离ρ(x,X_j),j=1,…,n,的上升秩序,把 X_1,…,X_n 重新排列成 X_((R)_1),…,     

下标降序法最近邻判别分析

§1 引言设(X_1,θ_1),…,(X_n,θ_n),(X,θ)为在R~d×{1,…,M}上取值的i.i.d.随机向量。问题是要利用X的观察值及历史样本(X_i,θ_i),i=1,…,n对类别变量θ进行判别。假定在R~d上给定了某一距离函数p(·,·)(比如欧氏距离等),那么可按照诸X_i与X的距离由小到大把诸X_i重新排列为X_(R1),X_(R2),…,X_(R_n,相应的θ_i也被排列为θ_(R1),θ_(R2),…,θ_(Rn)。若采用θ_(R1)来判别θ,这就是所谓的最近邻判别法。Derroye,Wagner,Fritz,陈希孺及白志     

可靠性增长试验中一类结构可靠度的统计分析

设X_(i1),X_(i2)……,X(in),是X_i的随机样本(i=1,2,…,m),且这些样本间是相互独立的,又设X_i～N(θ_i,δ_i~2)(θ_i与δ_i未知),且对某一常数c有P(X_1>c)≤P(X_2>c)≤…≤P(X_m>(?)基于上述样本,文中给出P(X_m>c)的点估计和在某种意义下最优的置信下限。     

非参数回归函数的基于截尾数据的估计

本文考虑截尾数据情况下非参数回归函数m(x)=E(Y|x)的估计。具体地讲,我们面对的是这样的数学模型:T是与(X,Y)独立的随机变量,我们观测到的不是Y本身,而是Z=min(Y,T)及δ=[Y≤T]。今有训练样本{(X_i,Z_i,δ_i)}_(i-1)及当前样本(X,z,δ),记ξ_i(·)=[z_i≥·], N~ (·)=sum from i=1 to n ξ_i(·), V_n(·)=multiply from i=1 to n{1 N~ (z_i)/2 N~ (z_i)}~[δ_i=_i<0], U_n(·)=sum from i=1 to n Wnt(x)ξ_i(·), 令 m_n(x)=integral from 0 to u_n U_n(y)|V_n(y)dy, 其中u_n=F_2~(-1)(n~(-a)),0<α<1/2为一实常数,F_2(·)=P(Y≥·)为Y的(右侧)分布函数。在权函数{W_(ni)(x)}_(i=1)~n及(X,Y,T)的分布函数满足一组条件下,我们证明了m_n(x)为m(x)的强相合估计,即:m_n(x)→m(x),a.s.(n→ ∞).     

经验Bayes估计渐近理论中的若干未决问题

这里介绍Singh在[3]中提出的关于经验Bayes估计的渐近理论的几个猜测,及某些有关问题。 设有绝对连续的一维指数分布族 dP_θ(x)=f_θ(x)dx=C(θ)~(θx)h(x)dx,θ∈ 为R~1上的一有限或无限区间。假定取平方损失L(a,θ)=(a-θ)~2θ有先验分布G.θ的Bayes估计记为d_G=d_G(x),其Bayes风险记为B(G). 设有历史样本X_1,…,X_n,当前样本X。按经验Bayes理论的基本假定,X_1,…,X_,(X,θ)相互独立,且每个X_i的分布与X的边缘分布相同。任一同时依赖于X_19,…,X_n和X的估计d_n=d_n(X_1,…,X_n,X)称为θ的一经验Bayes估计,其Bayes风险定义为     

用 k 近邻预测逼近 Bayes 预测

§1.方法与结果设(X,θ)是取值于 R~d×R~1上的随机变量。设已观察了 X 的值为 x,要利用 X 之值 x 去预测θ。假定δ(x)是一个预测函数,且引进了某种损失函数 L(θ,a)(用 a 去预测真值θ时有 L(θ,a)这么大的损失。则预测δ的风险定义为 R(δ)=E[L(θ,δ(X))]。若预测δ~*满足条件 R(δ~*)=(?) R(δ),其中 (?) 意义对一切可能的预测函数取 inf,则称δ~*为 Bayes 预测,R~*=R(δ~*)为 Bayes 风险。若(X,θ)的分布已知,原则上不难     

A Priori and A Posteriori: A Bootstrapping Relationship

The distinction between a priori and a posteriori knowledge has been the subject of an enormous amount of discussion, but the literature is biased against recognizing the intimate relationship between these forms of knowledge. For instance, it seems to be almost impossible to find a sample of pure a priori or a posteriori knowledge. In this paper, it will be suggested that distinguishing between a priori and a posteriori is more problematic than is often suggested, and that a priori and a posteriori resources are in fact used in parallel. We will define this relationship between a priori and a posteriori knowledge as the bootstrapping relationship. As we will see, this relationship gives us reasons to seek for an altogether novel definition of a priori and a posteriori knowledge. Specifically, we will have to analyse the relationship between a priori knowledge and a priori reasoning, and it will be suggested that the latter serves as a more promising starting point for the analysis of aprioricity. We will also analyse a number of examples from the natural sciences and consider the role of a priori reasoning in these examples. The focus of this paper is the analysis of the concepts of a priori and a posteriori knowledge rather than the epistemic domain of a posteriori and a priori justification.     

Subspaces and quotients of Banach spaces with shrinking unconditional bases

The main result is that a separable Banach space with the weak* unconditional tree property is isomorphic to a subspace as well as a quotient of a Banach space with a shrinking unconditional basis. A consequence of this is that a Banach space is isomorphic to a subspace of a space with a shrinking unconditional basis if and only if it is isomorphic to a quotient of a space with a shrinking unconditional basis, which solves a problem dating to the 1970s. The proof of the main result also yields that a uniformly convex space with the unconditional tree property is isomorphic to a subspace as well as a quotient of a uniformly convex space with an unconditional finite dimensional decomposition.     

A Barrier Function Method for the Nonconvex Quadratic Programming Problem with Box Constraints

In this paper a barrier function method is proposed for approximating a solution of the nonconvex quadratic programming problem with box constraints. The method attempts to produce a solution of good quality by following a path as the barrier parameter decreases from a sufficiently large positive number. For a given value of the barrier parameter, the method searches for a minimum point of the barrier function in a descent direction, which has a desired property that the box constraints are always satisfied automatically if the step length is a number between zero and one. When all the diagonal entries of the objective function are negative, the method converges to at least a local minimum point of the problem if it yields a local minimum point of the barrier function for a sequence of decreasing values of the barrier parameter with zero limit. Numerical results show that the method always generates a global or near global minimum point as the barrier parameter decreases at a sufficiently slow pace.     

Preservation of the existence of fixed and coincidence points under homotopy of mappings of ordered sets

In problems of topology and analysis, well-known theorem on the preservation by any continuous homotopy of the property of a mapping to have a fixed point and the property of a pair of mappings to have a coincidence point are extensively applied. Thus, for contraction mappings and some of their generalizations, Frigon’s results on the preservation of the property to have a fixed point by a homotopy of a special type are known. This paper presents theorems on the preservation by order homotopy of the property of a pair of mappings to have a coincidence point. As a corollary, conditions under which such a homotopy preserves the property of a mapping to have a fixed point are obtained.     

The Relative Maximum Genus of a Graph

A relative embedding of a graph in a surface with respect to a set of closed walks is one where each of the prescribed closed walks bounds a face of the embedding In the special case where the set of closed walks is empty, this amounts to the usual concept of a graph embedding. We present a formula for the maximum (orientable) genus of the surface on which a graph has a relative embedding with respect to a set of closed walks.     

Subgradient Projectors: Extensions,Theory, and Characterizations

Subgradient projectors play an important role in optimization and for solving convex feasibility problems. For every locally Lipschitz function, we can define a subgradient projector via generalized subgradients even if the function is not convex. The paper consists of three parts. In the first part, we study basic properties of subgradient projectors and give characterizations when a subgradient projector is a cutter, a local cutter, or a quasi-nonexpansive mapping. We present global and local convergence analyses of subgradent projectors. Many examples are provided to illustrate the theory. In the second part, we investigate the relationship between the subgradient projector of a prox-regular function and the subgradient projector of its Moreau envelope. We also characterize when a mapping is the subgradient projector of a convex function. In the third part, we focus on linearity properties of subgradient projectors. We show that, under appropriate conditions, a linear operator is a subgradient projector of a convex function if and only if it is a convex combination of the identity operator and a projection operator onto a subspace. In general, neither a convex combination nor a composition of subgradient projectors of convex functions is a subgradient projector of a convex function.     

A stabilized mixed finite element method for the biharmonic equation based on biorthogonal systems

We propose a stabilized finite element method for the approximation of the biharmonic equation with a clamped boundary condition. The mixed formulation of the biharmonic equation is obtained by introducing the gradient of the solution and a Lagrange multiplier as new unknowns. Working with a pair of bases forming a biorthogonal system, we can easily eliminate the gradient of the solution and the Lagrange multiplier from the saddle point system leading to a positive definite formulation. Using a superconvergence property of a gradient recovery operator, we prove an optimal a priori estimate for the finite element discretization for a class of meshes.     

Graphs with disjoint dominating and paired-dominating sets

A dominating set of a graph is a set of vertices such that every vertex not in the set is adjacent to a vertex in the set, while a paired-dominating set of a graph is a dominating set such that the subgraph induced by the dominating set contains a perfect matching. In this paper, we show that no minimum degree is sufficient to guarantee the existence of a disjoint dominating set and a paired-dominating set. However, we prove that the vertex set of every cubic graph can be partitioned into a dominating set and a paired-dominating set.     

Formulation of the problem of sonic boom by a maneuvering aerofoil as a one-parameter family of Cauchy problems

For the structure of a sonic boom produced by a simple aerofoil at a large distance from its source we take a physical model which consists of a leading shock (LS), a trailing shock (TS) and a one-parameter family of nonlinear wavefronts in between the two shocks. Then we develop a mathematical model and show that according to this model the LS is governed by a hyperbolic system of equations in conservation form and the system of equations governing the TS has a pair of complex eigenvalues. Similarly, we show that a nonlinear wavefront originating from a point on the front part of the aerofoil is governed by a hyperbolic system of conservation laws and that originating from a point on the rear part is governed by a system of conservation laws, which is elliptic. Consequently, we expect the geometry of the TS to be kink-free and topologically different from the geometry of the LS. In the last section we point out an evidence of kinks on the LS and kink-free TS from the numerical solution of the Euler’s equations by Inoue, Sakai and Nishida [5].     

Stability of a singularly perturbed problem

We prove the stability of the mixed problem for a system of telegraph equations under a perturbation of one of the boundary conditions by a sum of a singular perturbation (a small parameter multiplying the highest derivative) and a small regular perturbation. The solution of the problem consists of the current and voltage in a segment of a telegraph line. One of its ends is short-circuited, and a capacitor of small capacity, together with a nonlinear resistance whose volt-ampere characteristic is perturbed by a small term, is connected to the other end. We prove the convergence of the solution of the problem to the unique continuous piecewise continuously differentiable solution of the unperturbed problem bifurcating at some instant of time from its unique classical solution.