Covariance pooling is a feature pooling method with good classification accuracy. Because covariance features consist of second-order statistics, the scale of the feature elements are varied. Therefore, normalizing covariance features using a matrix square root affects the performance improvement. When pooling methods are applied to local features extracted from CNN models, the accuracy increases when the pooling function is back-propagatable and the feature-extraction model is learned in an end-to-end manner. Recently, the iterative polynomial approximation method for the matrix square root of a covariance feature was proposed, and resulted in a faster and more stable training than the methods based on singular-value decomposition. In this paper, we propose an extension of compact bilinear pooling, which is a compact approximation of the standard covariance feature, to the polynomials of the covariance feature. Subsequently, we apply the proposed approximation to the polynomial corresponding to the matrix square root to obtain a compact approximation for the square root of the covariance feature. Our method approximates a higher-dimensional polynomial of a covariance by the weighted sum of the approximate features corresponding to a pair of local features based on the similarity of the local features. We apply our method for standard fine-grained image recognition datasets and demonstrate that the proposed method shows comparable accuracy with fewer dimensions than the original feature.
The Matérn covariance function is a popular choice for prediction in spatial statistics and uncertainty quantification literature. A key benefit of the Matérn class is that it is possible to get precise control over the degree of mean-square differentiability of the random process. However, the Matérn class possesses exponentially decaying tails, and thus may not be suitable for modeling polynomially decaying dependence. This problem can be remedied using polynomial covariances; however one loses control over the degree of mean-square differentiability of corresponding processes, in that rando…
The problem of estimating the spectral density matrix $f(w)$ of a multivariate time series is revisited with special focus on the frequencies $w=0$ and $w=π$. Recognizing that the entries of the spectral density matrix at these two boundary points are real-valued, we propose a new estimator constructed from a local polynomial regression of the real portion of the multivariate periodogram. The case $w=0$ is of particular importance, since $f(0)$ is associated with the large-sample covariance matrix of the sample mean; hence, estimating $f(0)$ is crucial in order to conduct any sort of statistic…
Parsimonious Compactly Supported Covariance Models in the Gauss Hypergeometric Class: Identifiability, Reparameterizations, and Asymptotic Properties
Universidad Adolfo Ibáñez（阿道夫·伊巴涅斯大学）；Università Ca’ Foscari（威尼斯卡福斯卡里大学）；Universidad del Bío-Bío（比奥比奥大学）；Universidad de Santiago de Chile（智利圣地亚哥大学）；Universidad de Chile（智利大学）
Often machine learning and statistical models will attempt to describe the majority of the data. However, there may be situations where only a fraction of the data can be fit well by a linear regression model. Here, we are interested in a case where such inliers can be identified by a Disjunctive Normal Form (DNF) formula. We give a polynomial time algorithm for the conditional linear regression task, which identifies a DNF condition together with the linear predictor on the corresponding portion of the data. In this work, we improve on previous algorithms by removing a requirement that the co…
Using Bernstein polynomial approximations, we prove the central limit theorem for linear spectral statistics of sample covariance matrices, indexed by a set of functions with continuous fourth order derivatives on an open interval including $[(1-\sqrt{y})^2,(1+\sqrt{y})^2]$, the support of the Marucenko--Pastur law. We also derive the explicit expressions for asymptotic mean and covariance functions.
We observe a sample of $n$ independent $p$-dimensional Gaussian vectors with Toeplitz covariance matrix $ Σ= [σ_{|i-j|}]_{1 \leq i,j \leq p}$ and $σ_0=1$. We consider the problem of testing the hypothesis that $Σ$ is the identity matrix asymptotically when $n \to \infty$ and $p \to \infty$. We suppose that the covariances $σ_k$ decrease either polynomially ($\sum_{k \geq 1} k^{2α} σ^2_{k} \leq L$ for $ α>1/4$ and $L>0$) or exponentially ($\sum_{k \geq 1} e^{2Ak} σ^2_{k} \leq L$ for $ A,L>0$). We consider a test procedure based on a weighted U-statistic of order 2, with optimal weights chosen a…
Under the high-dimensional setting that data dimension and sample size tend to infinity proportionally, we derive the central limit theorem (CLT) for linear spectral statistics (LSS) of large-dimensional sample covariance matrix. Different from existing literature, our results do not require the assumption that the population covariance matrices are bounded. Moreover, many common kernel functions in the real data such as logarithmic functions and polynomial functions are allowed in this paper. In our model, the number of spiked eigenvalues can be fixed or tend to infinity. One salient feature …
In this paper, the uniformly asymptotic normality for sample quantiles of associated random variables is investigated under some conditions on the decay of the covariances. We obtain the rate of normal approximation of order $O(n^{-1/2}\log^2 n)$ if the covariances decrease exponentially to $0$. The best rate is shown as $O(n^{-1/3})$ under a polynomial decay of the covariances.
Directed graphical models specify noisy functional relationships among a collection of random variables. In the Gaussian case, each such model corresponds to a semi-algebraic set of positive definite covariance matrices. The set is given via parametrization, and much work has gone into obtaining an implicit description in terms of polynomial (in-)equalities. Implicit descriptions shed light on problems such as parameter identification, model equivalence, and constraint-based statistical inference. For models given by directed acyclic graphs, which represent settings where all relevant variable…
Spatial-sign covariance matrix (SSCM) is an important substitute of sample covariance matrix (SCM) in robust statistics. This paper investigates the SSCM on its asymptotic spectral behaviors under high-dimensional elliptical populations, where both the dimension $p$ of observations and the sample size $n$ tend to infinity with their ratio $p/n\to c\in (0, \infty)$. The empirical spectral distribution of this nonparametric scatter matrix is shown to converge in distribution to a generalized Marčenko-Pastur law. Beyond this, a new central limit theorem (CLT) for general linear spectral statistic…
This work provides a unified analysis of the properties of the sample covariance matrix $Σ_n$ over the class of $p\times p$ population covariance matrices $Σ$ of reduced effective rank $r_e(Σ)$. This class includes scaled factor models and covariance matrices with decaying spectrum. We consider $r_e(Σ)$ as a measure of matrix complexity, and obtain sharp minimax rates on the operator and Frobenius norm of $Σ_n-Σ$, as a function of $r_e(Σ)$ and $\|Σ\|_2$, the operator norm of $Σ$. With guidelines offered by the optimal rates, we define classes of matrices of reduced effective rank over which $Σ…
We study estimation and prediction of Gaussian random fields with covariance models belonging to the generalized Wendland (GW) class, under fixed domain asymptotics. As the Matérn case, this class allows a continuous parameterization of smoothness of the underlying Gaussian random field, being additionally compactly supported. The paper is divided into two parts: First, we characterize the equivalence of two Gaussian measures with GW covariance function, and we provide sufficient conditions for the equivalence of two Gaussian measures with Matérn and GW covariance functions. We elucidate the c…
Istituto Italiano di Tecnologia（意大利技术研究院）；University of Verona（维罗纳大学）
This paper presents a novel framework for visual object recognition using infinite-dimensional covariance operators of input features in the paradigm of kernel methods on infinite-dimensional Riemannian manifolds. Our formulation provides in particular a rich representation of image features by exploiting their non-linear correlations. Theoretically, we provide a finite-dimensional approximation of the Log-Hilbert-Schmidt (Log-HS) distance between covariance operators that is scalable to large datasets, while maintaining an effective discriminating capability. This allows us to efficiently app…
Facial expressions recognition (FER) of 3D face scans has received a significant amount of attention in recent years. Most of the facial expression recognition methods have been proposed using mainly 2D images. These methods suffer from several issues like illumination changes and pose variations. Moreover, 2D mapping from 3D images may lack some geometric and topological characteristics of the face. Hence, to overcome this problem, a multi-modal 2D + 3D feature-based method is proposed. We extract shallow features from the 3D images, and deep features using Convolutional Neural Networks (CNN)…
