arXiv research

CASP improves portfolio optimization by considering asset covariance.

Statistical modeling of spatiotemporal phenomena often requires selecting a covariance matrix from a covariance class. Yet standard parametric covariance families can be insufficiently flexible for practical applications, while non-parametric approaches may not easily allow certain kinds of prior knowledge to be incorp…

New method clusters high-dimensional data with anisotropic noise.

This paper focuses on the estimation of the sample covariance matrix from low-dimensional random projections of data known as compressive measurements. In particular, we present an unbiased estimator to extract the covariance structure from compressive measurements obtained by a general class of random projection matri…

Study variance-optimal hedging of forward curve derivatives under stochastic volatility.

New method selects variables for GP regression using sparse projection.

This paper improves computational efficiency in kernel ridge regression under covariate shift.

Mixture models are a standard approach to dealing with heterogeneous data with non-i.i.d. structure. However, when the dimension $p$ is large relative to sample size $n$ and where either or both of means and covariances/graphical models may differ between the latent groups, mixture models face statistical and computati…

Study projective representations of infinite-dimensional Hilbert-Lie groups.

Here, a non-linear analysis method is applied rather than classical one to study projective changes of Finsler metrics. More intuitively, a projectively invariant pseudo-distance is introduced and characterized with respect to the Ricci tensor and its covariant derivatives.

Sharp-SSL uses random projections to identify important variables for semi-supervised learning.

New method generates synthetic time series paths with more flexibility.

Dynamic treatment effects estimated over time using covariate balancing.

P.Lecomte has proposed to take into account the covariant derivatives used to build ordering prescriptions for the naturality of transformation properties and has conjectured that there exists an natural ordering prescription for differential operators of any orders between density bundles which in addition is invarian…

We search for Riemannian metrics whose Levi-Civita connection belongs to a given projective class. Following Sinjukov and Mikes, we show that such metrics correspond precisely to suitably positive solutions of a certain projectively invariant finite-type linear system of partial differential equations. Prolonging this …

Paper proposes a novel approach to density ratio estimation using projection pursuit.

Estimation of low-rank matrices is of significant interest in a range of contemporary applications. In this paper, we introduce a rank-one projection model for low-rank matrix recovery and propose a constrained nuclear norm minimization method for stable recovery of low-rank matrices in the noisy case. The procedure is…

PCA outperforms random projections in retaining second order signals from latent groups.

Latent space models are effective tools for statistical modeling and exploration of network data. These models can effectively model real world network characteristics such as degree heterogeneity, transitivity, homophily, etc. Due to their close connection to generalized linear models, it is also natural to incorporat…

Random projections offer an appealing and flexible approach to a wide range of large-scale statistical problems. They are particularly useful in high-dimensional settings, where we have many covariates recorded for each observation. In classification problems there are two general techniques using random projections. T…

The notion of a Kähler structure for a differential calculus was recently introduced by the second author as a framework in which to study the noncommutative geometry of the quantum flag manifolds. It was subsequently shown that any covariant positive definite Kähler structure has a canonically associated triple satisf…

A new GNN architecture called coVariance neural network (VNN) improves stability and transferability of covariance matrix analysis.

Gaussian Markov random fields (GMRFs) are useful in a broad range of applications. In this paper we tackle the problem of learning a sparse GMRF in a high-dimensional space. Our approach uses the l1-norm as a regularization on the inverse covariance matrix. We utilize a novel projected gradient method, which is faster …

In the classical Gaussian SVM classification we use the feature space projection transforming points to normal distributions with fixed covariance matrices (identity in the standard RBF and the covariance of the whole dataset in Mahalanobis RBF). In this paper we add additional information to Gaussian SVM by considerin…

Fast algorithm recovers principal eigenvector from noisy matrices.

Inflating the minimum norm interpolator improves linear regression generalization error.

This is the second in a series of papers on natural modification of the normal tractor connection in a parabolic geometry, which naturally prolongs an underlying overdetermined system of invariant differential equations. We give a short review of the general procedure developed in [5] and then compute the prolongation …

Study mini-batch SGD noise and its limits, proving complexity guarantees.

Geodesic sprays on Finsler manifolds studied with covariant coefficients.

FSPA bypasses eigenvalue estimation for quantum PCA, achieving optimal complexity and robustness.

We discuss the geometric foundation behind the use of stochastic processes in the frame bundle of a smooth manifold to build stochastic models with applications in statistical analysis of non-linear data. The transition densities for the projection to the manifold of Brownian motions developed in the frame bundle lead …

We formalize causal separation in portfolio theory, deriving a closed-form projected Markowitz solution.

We compute the eigenvalues with multiplicities of the Lichnerowicz Laplacian acting on the space of complex symmetric covariant tensor fields on the complex projective space $P^n(\comp)$. The spaces of symmetric eigentensors are explicitly given.

Holomorphic structures on quantum flag manifolds uniquely defined.

CSTs improve stability in covariance spectrum analysis without training.

We derive an efficient method to perform clustering of nodes in Gaussian graphical models directly from sample data. Nodes are clustered based on the similarity of their network neighborhoods, with edge weights defined by partial correlations. In the limited-data scenario, where the covariance matrix would be rank-defi…

New method learns decisions from collective preferences without individual covariates.

Nyström subsampling with Tikhonov regularization for covariate shift adaptation under misspecified case

Exterior differential forms with values in the (Kostant's) symplectic spinor bundle on a manifold with a given metaplectic structure are decomposed into invariant subspaces. Projections to these invariant subspaces of a covariant derivative associated to a torsion-free symplectic connection are described.

We sketch our recent application of a non-commutative version of the Cartan `moving-frame' formalism to the quantum Euclidean space $R^N_q$, the space which is covariant under the action of the quantum group $SO_q(N)$. For each of the two covariant differential calculi over $R^N_q$ based on the $R$-matrix formalism, we…

Develops MGQDA for multi-group classification with theoretical guarantees and practical applications.

The paper examines extreme value statistics of high-dimensional sample covariances, with applications in finance and image analysis.

The problem of image restoration in cryo-EM entails correcting for the effects of the Contrast Transfer Function (CTF) and noise. Popular methods for image restoration include `phase flipping', which corrects only for the Fourier phases but not amplitudes, and Wiener filtering, which requires the spectral signal to noi…

A new model for dynamic covariance recovery in neuroimaging data.

The paper shows how sketching data can simplify regression inference even when errors are heteroskedastic.

A new method tests variable significance without assuming model correctness.

A new method routes EEG covariance matrices across domains using adaptive subspace selection.

Develops a universal Hermitian projective calculus for complex hyperbolic two-space