Searcharxiv⌕ Search

arXiv subjects

Van Vu

Publications and source records attributed to Van Vu.

At least 19 recordsLinked to original sources
arXiv2026

Estimating eigenvectors and eigenspaces of covariance matrices: Optimal Bounds and Conditions for Consistency

Let $X = [ \xi_1, \,\, \xi_2,...\,\, ,\xi_d]^\top$ be a zero-mean random vector of large dimension $d$ ($d \rightarrow \infty$) with (hidden) covariance matrix $M = (m_{ij})_{1 \leq i, j \leq d},$ where $m_{ij} = m_{ji} = \textbf{Cov}(\xi_i, \xi_j).$ Let $X_1, X_2, \dots, X_n$ be $n$ iid samples of $X$. Consider the sample covariance matrix $$\textstyle \tilde{M} := \frac{1}{n} \sum_{i=1}^{n} X_i X_i^\top.$$ In practice, one frequently uses the eigenvectors and eigenspaces of $\tilde M$ as estimators for those of $M$. A central task is to provide an error analysis for these estimators. In this paper, we provide an optimal error analysis, obtaining upper and lower bounds of matching order of magnitude, for a wide range of parameters $d$ and $n$, under mild assumptions on $M$. As corollaries, we obtain new necessary and sufficient conditions for the consistency of the estimators. In these conditions, we only require the number of samples $n$ to depend linearly on the effective rank of $M$, which can be much smaller than the dimension $d$.

math.ST↗
arXiv2026

Eigenvalue stability and new perturbation bounds for the extremal eigenvalues of a matrix

Let $A$ be a full ranked $ n\times n$ matrix, with singular values $\sigma_1 (A) \ge \dots \ge \sigma_n (A) >0$. The condition number $\kappa(A):= \sigma_1(A)/\sigma_n(A)=\|A\|\cdot \|A\|^{-1}$ is a key parameter in the analysis of algorithms taking $A$ as input. In practice, matrices (representing real data) are often perturbed by noise. Technically speaking, the real input would be a noisy variant $\tilde A =A +E$ of $A$, where $E$ represents the noise. The condition number $\kappa (\tilde A)$ will be used instead of $\kappa (A)$. Thus, it is of importance to measure the impact of noise on the condition number. In this paper, we focus on the case when the noise is random. We introduce the notion of regional stability, via which we design a new framework to estimate the perturbation of the extremal singular values and the condition number of a matrix. Our framework allows us to bound the perturbation of singular values through the perturbation of singular spaces. We then bound the latter using a novel contour analysis argument, which, as a co-product, provides an improved version of the classical Davis-Kahan theorem in many settings. Our new estimates concerning the least singular value $\sigma_n(A)$ complement well-known results in this area, and are more favorable in the case when the ground matrix $A$ is large compared to the noise matrix $E$.

math.NA↗
arXiv2026

Anti-concentration with respect to random permutations

Classical anti-concentration results focus on the random sum $S := \sum _{i=1}^n \xi_i v_i$, where $\xi_i$ are independent random variables and $v_i$ are real numbers. In this paper, we prove new concentration results concerning the random sum $S := \sum_{i=1}^n w_{\pi _i } v_i $, where $w_i , v_i$ are real numbers and $\pi$ is a random permutation.

math.PR↗
arXiv2025

Matrices perturbed by random noise: The accuracy of low-rank approximation

Let $A$ be an $m \times n$ matrix with rank $r$ and singular value decomposition $A = \sum _{i=1}^r \sigma_i u_i v_i^\top, $ where the $\sigma_i$ are its singular values, ordered decreasingly, and $u_i, v_i$ are the corresponding left and right singular vectors. For an integer $1 \le p \le r$, $A_p := \sum_{i=1}^p \sigma_i u_i v_i^\top$ is the best rank-$p$ approximation of $A$. In practice, one often chooses $p$ to be small, leading to the commonly used phrase ``low-rank approximation''. For a large data matrix $A$, one typically computes a rank-$p$ approximation $A_p$ for a suitably chosen small $p$, stores $A_p$, and uses it as input for further computations. The reduced dimension of $A_p$ enables faster computations and significant data compression. In practice, noise is inevitable. We often have access only to noisy data $\tilde A = A + E$, where $E$ represents the noise. Consequently, the low-rank approximation used as input in many downstream tasks is $\tilde A_p$, the best rank-$p$ approximation of $\tilde A$, rather than $A_p$. Therefore, it is natural and important to estimate the error $ \| \tilde A_p - A_p \|$. This error plays a critical role in assessing the accuracy of downstream procedures involving low-rank approximations of noisy inputs. The standard way to estimate this error is to use the Eckart--Young--Mirsky identity, which provides $$\| \tilde A_p -A_p\| = O(\sigma_{p+1} + \|E\|).$$ A situation that occurs frequently in applications is that noise is random and $A$ has relatively low rank. In this situation, $\sigma_{p+1}$ often dominates $\|E \|$ by a large factor. Our main results in this paper show that in this situation, it is possible to improve the Eckart--Young--Mirsky bound by removing the term $\sigma_{p+1}$ completely or replacing it by a much smaller quantity.

math.NA↗
arXiv2025

Davis-Kahan Theorem under a moderate gap condition

The classical Davis-Kahan theorem provides an efficient bound on the perturbation of eigenspaces of a matrix under a large (eigenvalue) gap condition. In this paper, we consider the case when the gap is moderate. Using a bootstrapping argument, we obtain a new bound which is efficient when the perturbation matrix is uncorrelated to the ground matrix. We believe that this bound is sharp up to a logarithmic term.

math.NA↗
arXiv2025

Fast exact recovery of noisy matrix from few entries: the infinity norm approach

The matrix recovery (completion) problem, a central problem in data science and theoretical computer science, is to recover a matrix $A$ from a relatively small sample of entries. While such a task is impossible in general, it has been shown that one can recover $A$ exactly in polynomial time, with high probability, from a random subset of entries, under three (basic and necessary) assumptions: (1) the rank of $A$ is very small compared to its dimensions (low rank), (2) $A$ has delocalized singular vectors (incoherence), and (3) the sample size is sufficiently large. There are many different algorithms for the task, including convex optimization by Candes, Tao and Recht (2009), alternating projection by Hardt and Wooters (2014) and low rank approximation with gradient descent by Keshavan, Montanari and Oh (2009, 2010). In applications, it is more realistic to assume that data is noisy. In this case, these approaches provide an approximate recovery with small root mean square error. However, it is hard to transform such an approximate recovery to an exact one. Recently, results by Abbe et al. (2017) and Bhardwaj et al. (2023) concerning approximation in the infinity norm showed that we can achieve exact recovery even in the noisy case, given that the ground matrix has bounded precision. Beyond the three basic assumptions above, they required either the condition number of $A$ is small (Abbe et al.) or the gap between consecutive singular values is large (Bhardwaj et al.). In this paper, we remove these extra spectral assumptions. As a result, we obtain a simple algorithm for exact recovery in the noisy case, under only the three basic assumptions. This is the first such algorithm. To analyse this algorithm, we introduce a contour integration argument which is totally different from all previous methods and may be of independent interest.

math.ST↗
arXiv2024

New matrix perturbation bounds with relative norm: Perturbation of eigenspaces

Matrix perturbation bounds (such as Weyl and Davis-Kahan) are used abundantly in many areas of mathematics and data science. Many bounds (such as the above two) involve the spectral norm of the noise matrix and are sharp in worst case analysis. In order to refine these classical bounds, we introduce a new parameter, which we refer to as the relative norm. This parameter measures the strength of the action of the noise matrix on the relevant eigenvectors of the ground matrix. It has turned out that in a number of situations, we can use the relative norm as a replacement for the spectral norm. This has led to a number of notable improvements under certain sets of assumptions, which are frequently met in practice. For instance, our new results apply very well in the case when the noise matrix is random. For the purpose of our study, we introduce a new method of analysis, which combines the classical contour integral argument with new (combinatorial) ideas. This method is robust and of independent interest. In the current paper, we focus on the perturbation of eigenspaces (Davis-Kahan type results). Perturbation bounds for eigenspaces are essential in statistics and theoretical computer science, and thus deserve a special treatment. Furthermore, this will lay the ground for the more technical treatment of general matrix functionals, which appears in a future paper.

math.SP↗
arXiv2024

Matrix perturbation bounds via contour bootstrapping

Matrix perturbation bounds play an essential role in the design and analysis of spectral algorithms. In this paper, we use a "contour bootstrapping" argument to derive several new perturbation bounds. As applications, we discuss new bounds on the error occurring when one uses matrix sparsification to speed up the computation of spectral parameters. Another potential application is the estimation of the trade-off in computing with privacy.

math.NA↗
arXiv2023

Matrix Perturbation: Davis-Kahan in the Infinity Norm

Perturbation theory is developed to analyze the impact of noise on data and has been an essential part of numerical analysis. Recently, it has played an important role in designing and analyzing matrix algorithms. One of the most useful tools in this subject, the Davis-Kahan sine theorem, provides an $\ell_2$ error bound on the perturbation of the leading singular vectors (and spaces). We focus on the case when the signal matrix has low rank and the perturbation is random, which occurs often in practice. In an earlier paper, O'Rourke, Wang, and the second author showed that in this case, one can obtain an improved theorem. In particular, the noise-to-gap ratio condition in the original setting can be weakened considerably. In the current paper, we develop an infinity norm version of the O'Rourke-Vu-Wang result. The key ideas in the proof are a new bootstrapping argument and the so-called iterative leave-one-out method, which may be of independent interest. Applying the new bounds, we develop new, simple, and quick algorithms for several well-known problems, such as finding hidden partitions and matrix completion. The core of these new algorithms is the fact that one is now able to quickly approximate certain key objects in the infinity norm, which has critical advantages over approximations in the $\ell_2$ norm, Frobenius norm, or spectral norm.

math.PR↗
arXiv2023

The "Power of Few" Phenomenon: The Sparse Case

The "majority dynamics" process on a social network begins with an initial phase, where the individuals are split into two competing parties, Red and Blue. Every day, everyone updates their affiliation to match the majority among those of their friends. While studying this process on Erdos-Renyi G(n, p) random graph (with constant density), the authors discovered the "Power of Few" phenomenon, showing that a very small advantage to one side already guarantees that everybody will unanimously join that side after just a few days with overwhelming probability. For example, when p = 1/2, then 10 extra members guarantee this unanimity with a 90% chance, regardless of the value of n. In this paper, we study this phenomenon for sparse random graphs. It is clear that below the connectivity threshold, the phenomenon ceases to hold, as the isolated vertices never change their colors. We show that it holds for every density above the threshold. To make the process more realistic, we also assume that individuals can randomly activate their accounts to post their opinions and observe their neighbors (just as we do on social media). We prove that the phenomenon is robust under this assumption.

math.CO↗
arXiv2022

VinDr-Mammo: A large-scale benchmark dataset for computer-aided diagnosis in full-field digital mammography

Mammography, or breast X-ray, is the most widely used imaging modality to detect cancer and other breast diseases. Recent studies have shown that deep learning-based computer-assisted detection and diagnosis (CADe or CADx) tools have been developed to support physicians and improve the accuracy of interpreting mammography. However, most published datasets of mammography are either limited on sample size or digitalized from screen-film mammography (SFM), hindering the development of CADe and CADx tools which are developed based on full-field digital mammography (FFDM). To overcome this challenge, we introduce VinDr-Mammo - a new benchmark dataset of FFDM for detecting and diagnosing breast cancer and other diseases in mammography. The dataset consists of 5,000 mammography exams, each of which has four standard views and is double read with disagreement (if any) being resolved by arbitration. It is created for the assessment of Breast Imaging Reporting and Data System (BI-RADS) and density at the breast level. In addition, the dataset also provides the category, location, and BI-RADS assessment of non-benign findings. We make VinDr-Mammo publicly available on PhysioNet as a new imaging resource to promote advances in developing CADe and CADx tools for breast cancer screening.

eess.IV↗
arXiv2021

VinDr-SpineXR: A deep learning framework for spinal lesions detection and classification from radiographs

Radiographs are used as the most important imaging tool for identifying spine anomalies in clinical practice. The evaluation of spinal bone lesions, however, is a challenging task for radiologists. This work aims at developing and evaluating a deep learning-based framework, named VinDr-SpineXR, for the classification and localization of abnormalities from spine X-rays. First, we build a large dataset, comprising 10,468 spine X-ray images from 5,000 studies, each of which is manually annotated by an experienced radiologist with bounding boxes around abnormal findings in 13 categories. Using this dataset, we then train a deep learning classifier to determine whether a spine scan is abnormal and a detector to localize 7 crucial findings amongst the total 13. The VinDr-SpineXR is evaluated on a test set of 2,078 images from 1,000 studies, which is kept separate from the training set. It demonstrates an area under the receiver operating characteristic curve (AUROC) of 88.61% (95% CI 87.19%, 90.02%) for the image-level classification task and a mean average precision (mAP@0.5) of 33.56% for the lesion-level localization task. These results serve as a proof of concept and set a baseline for future research in this direction. To encourage advances, the dataset, codes, and trained deep learning models are made publicly available.

eess.IV↗
arXiv2020

VinDr-CXR: An open dataset of chest X-rays with radiologist's annotations

Most of the existing chest X-ray datasets include labels from a list of findings without specifying their locations on the radiographs. This limits the development of machine learning algorithms for the detection and localization of chest abnormalities. In this work, we describe a dataset of more than 100,000 chest X-ray scans that were retrospectively collected from two major hospitals in Vietnam. Out of this raw data, we release 18,000 images that were manually annotated by a total of 17 experienced radiologists with 22 local labels of rectangles surrounding abnormalities and 6 global labels of suspected diseases. The released dataset is divided into a training set of 15,000 and a test set of 3,000. Each scan in the training set was independently labeled by 3 radiologists, while each scan in the test set was labeled by the consensus of 5 radiologists. We designed and built a labeling platform for DICOM images to facilitate these annotation procedures. All images are made publicly available (https://www.physionet.org/content/vindr-cxr/1.0.0/) in DICOM format along with the labels of both the training set and the test set.

eess.IV↗
arXiv2020

Random orthonormal polynomials: local universality and expected number of real roots

We consider random orthonormal polynomials $$ F_{n}(x)=\sum_{i=0}^{n}\xi_{i}p_{i}(x), $$ where $\xi_{0}$, \dots, $\xi_{n}$ are independent random variables with zero mean, unit variance and uniformly bounded $(2+\ep)$ moments, and $(p_n)_{n=0}^{\infty}$ is the system of orthonormal polynomials with respect to a fixed compactly supported measure on the real line. Under mild technical assumptions satisfied by many classes of classical polynomial systems, we establish universality for the leading asymptotics of the average number of real roots of $F_n$, both globally and locally. Prior to this paper, these results were known only for random orthonormal polynomials with Gaussian coefficients \cite{lubinsky2016linear} using the Kac-Rice formula, a method that does not extend to the generality of our paper.

math.PR↗
arXiv2020

Reaching a Consensus on Random Networks: The Power of Few

A community of $n$ individuals splits into two camps, Red and Blue. The individuals are connected by a social network, which influences their colors. Everyday, each person changes his/her color according to the majority among his/her neighbors. Red (Blue) wins if everyone in the community becomes Red (Blue) at some point. We study this process when the underlying network is the random Erdos-Renyi graph $G(n, p)$. With a balanced initial state ($n/2$ person in each camp), it is clear that each color wins with the same probability. Our study reveals that for any constants $p$ and $\varepsilon$, there is a constant $C$ such that if one camp has $n/2 +C$ individuals, then it wins with probability at least $1 - \varepsilon$. The surprising key fact here is that $C$ does not depend on $n$, the population of the community. When $p=1/2$ and $\varepsilon =.1$, one can set $C$ as small as 6. If the aim of the process is to choose a candidate, then this means it takes only $6$ "defectors" to win an election unanimously with overwhelming odd.

math.CO↗
arXiv2019

Random matrix products: Universality and least singular values

We establish, under a moment matching hypothesis, the local universality of the correlation functions associated with products of $M$ independent iid random matrices, as $M$ is fixed, and the sizes of the matrices tend to infinity. This generalizes an earlier result of Tao and the third author for the case $M=1$. We also prove Gaussian limits for the centered linear spectral statistics of products of $M$ independent iid random matrices. This is done in two steps. First, we establish the result for product random matrices with Gaussian entries, and then extend to the general case of non-Gaussian entries by another moment matching argument. Prior to our result, Gaussian limits were known only for the case $M=1$. In a similar fashion, we establish Gaussian limits for the centered linear spectral statistics of products of independent truncated random unitary matrices. In both cases, we are able to obtain explicit expressions for the limiting variances. The main difficulty in our study is that the entries of the product matrix are no longer independent. Our key technical lemma is a lower bound on the least singular value of the translated linearization matrix associated with the product of $M$ normalized independent random matrices with independent and identically distributed subgaussian entries. This lemma is of independent interest.

math.PR↗
arXiv2019

Random polynomials: central limit theorems for the real roots

The number of real roots has been a central subject in the theory of random polynomials and random functions since the fundamental papers of Littlewood-Offord and Kac in the 1940s. The main task here is to determine the limiting distribution of this random variable. In 1974, Maslova famously proved a central limit theorem (CLT) for the number of real roots of Kac polynomials. It has remained the only limiting theorem available for the number of real roots for more than four decades. In this paper, using a new approach, we derive a general CLT for the number of real roots of a large class of random polynomials with coefficients growing polynomially. Our result both generalizes and strengthens Maslova's theorem.

math.PR↗