arXiv 1709.014302017-09math.STmath.PRstat.TH

A class of non-parametric deformed exponential statistical models

Luigi Montrucchio, Giovanni Pistone

备注：: Revised submission to proceedings GSI 2017 Conference, Paris

英文摘要：

We study the class on non-parametric deformed statistical models where the deformed exponential has linear growth at infinity and is sub-exponential at zero. This class generalizes the class introduced by N.J.~Newton. We discuss the convexity and regularity of the normalization operator, the form of the deformed statistical divergences and their convex duality, the properties of the escort densities, and the affine manifold structure of the statistical bundle.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

阅读原文

阅读 PDF ↗

arXiv 页面 ↗HTML 原文 ↗

本文内容

英文摘要相关论文

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

继续探索

相关论文

变形指数族的边缘多面体

Marginal Polytope of a Deformed Exponential Family

一种变形对数函数$q$ -logarithm自C.察利斯J. Naudts提出了一种称为$ϕ$ -对数的推广，并导出了$ϕ$ -指数族的基本性质。本文研究了有限状态空间中的边缘多面体的相关概念。

2011-12-21math.STstat.TH

非线性非平衡二项回归与变形指数族

Infinitely imbalanced binomial regression and deformed exponential families

如果二进制响应趋于无穷不平衡，则已知逻辑斯谛回归模型收敛到泊松点过程模型。在本文中，这一现象是普遍的，在一个广泛的一类链接函数的二项式回归。证明依赖于极值理论。对于logit、probit和互补log-log连接函数，点过程的强度测度成为指数族。对于其他一些链接函数，变形指数族出现。提出了Poisson点过程模型的惩罚极大似然估计。

2013-04-21math.STstat.MLstat.TH

$κ$ -从几何观点看指数模型

$κ$-exponential models from the geometrical viewpoint

我们讨论使用Kaniadakis的$κ$ -指数的建设的统计流形上的Lebesgue空间的实随机变量。考虑了变形指数模型的一些代数特征。为每个严格正密度定义一个图表;参考概率的适当邻域中的每个其他严格正密度由中心$\Kln$似然表示

2009-03-11math.PRmath.STstat.TH

参数估计中的惩罚指数风险界

A penalized exponential risk bound in parametric estimation

本文提出了一种新的统一方法来研究拟似然法参数估计的精度。该方法的重要特征是：（1）基础模型{不假设是参数的}。(2)不需要参数可识别性的条件。参数集可以是无界的。(3)模型假设是相当一般的，没有具体的结构性假设，如独立或弱依赖的观察。对模型施加的条件非常温和，在特定应用中可以很容易地检查。(4)所建立的风险界是{非渐近}的，并适用于大，中，小样本。(5)主要结果是准极大似然估计的浓度属性，给出了所考虑的估计偏离“真”点的小邻域的概率的非渐近指数界。在轻度正则性条件下的标准情况下，通常的一致性和速率结果可以很容易地从已建立的风险界得到推论。以广义线性模型和单指标模型为例说明了方法和结果。

2009-03-10math.STstat.TH

New power-law tailed distributions emerging in $κ$-statistics

Over the last two decades, it has been argued that the Lorentz transformation mechanism, which imposes the generalization of Newton's classical mechanics into Einstein's special relativity, implies a generalization, or deformation, of the ordinary statistical mechanics. The exponential function, which defines the Boltzmann's factor, emerges properly deformed within this formalism. Starting from this, so-called $κ$-deformed exponential function, we introduce new classes of statistical distributions emerging as the $κ$-deformed version of already known distribution as the Generalized Gamma, Weib…

2022-03-03cond-mat.stat-mechmath-phmath.MP

Translations in the exponential Orlicz space with Gaussian weight

We study the continuity of space translations on non-parametric exponential families based on the exponential Orlicz space with Gaussian reference density.

2017-08-31math.STstat.TH

更多相关论文收起更多论文

通过SPD几何在非参数指数统计流形上的可处理伪度量

A Tractable Pseudo-Metric on Non-Parametric Exponential Statistical Manifolds via SPD Geometry

Indian Institute of Space Science and Technology (IIST)（印度空间科学与技术学院）

研究非参数统计流形上概率分布距离难计算问题，提出两阶段框架，先投影到有限维参数指数族，再嵌入对称正定矩阵流形，得到可计算伪度量及仿射不变检验统计量，应用于两样本检验，无需连续调整参数。

2026-07-14math.DGmath.STstat.TH

线性参数模型下混合有序与指数族因果有向无环图的可识别性

On the Identifiability of Mixed Ordinal and Exponential Family Causal DAGs under Linear Parametric Models

University of Southern California（南加州大学）

本文研究线性参数模型中混合有序与指数族节点间因果边的方向可识别性，证明在有序节点至少三类别且指数族节点至少三支撑点条件下，边方向可由联合分布唯一识别，并通过实验验证了理论。

2026-09-17cs.LGmath.STstat.ME

跳跃扩散过程漂移的非参数自适应估计

Non-parametric adaptive estimation of the drift for a jump diffusion process

本文考虑在离散时刻t= 0，Delta，.观察到的跳跃扩散过程（X_t），n Δ。采样间隔Delta趋于0，nDelta趋于无穷大。我们假设（X_t）是遍历的，严格平稳的，指数β-混合的.我们使用惩罚最小二乘方法来计算漂移函数b的两个自适应估计。我们提供了两个估计的风险的界限。

2013-09-26math.STstat.TH

Distributionally Robust Parametric Maximum Likelihood Estimation

Stanford University（斯坦福大学）；University of Cyprus（塞浦路斯大学）

We consider the parameter estimation problem of a probabilistic generative model prescribed using a natural exponential family of distributions. For this problem, the typical maximum likelihood estimator usually overfits under limited training sample size, is sensitive to noise and may perform poorly on downstream predictive tasks. To mitigate these issues, we propose a distributionally robust maximum likelihood estimator that minimizes the worst-case expected log-loss uniformly over a parametric Kullback-Leibler ball around a parametric nominal distribution. Leveraging the analytical expressi…

2020-10-11cs.LGmath.STstat.ML

Minimum $L^q$-distance estimators for non-normalized parametric models

We propose and investigate a new estimation method for the parameters of models consisting of smooth density functions on the positive half axis. The procedure is based on a recently introduced characterization result for the respective probability distributions, and is to be classified as a minimum distance estimator, incorporating as a distance function the $L^q$-norm. Throughout, we deal rigorously with issues of existence and measurability of these implicitly defined estimators. Moreover, we provide consistency results in a common asymptotic setting, and compare our new method with classic…

2020-03-13math.STstat.MEstat.TH

Exponential increase of the power of the independence and homogeneity chi-square tests with auxiliary information

This paper is an extension of the work about the exponential increase of the power of two non-parametric tests: the $ Z $-test and the chi-square goodness-of-fit test. Subject to having auxiliary information, it is possible to improve exponentially relative to the size of the sample the power of the famous chi-square tests of independence and homogeneity. Improving the power of these statistical tests by using auxiliary information makes it possible either to reduce the probability of accepting the null hypothesis under the alternative hypothesis, or to reduce the size of the sample necessary …

2021-09-01math.STstat.APstat.CO