The Lasso is a method for high-dimensional regression, which is now commonly used when the number of covariates $p$ is of the same order or larger than the number of observations $n$. Classical asymptotic normality theory does not apply to this model due to two fundamental reasons: $(1)$ The regularized risk is non-smooth; $(2)$ The distance between the estimator $\widehat{\boldsymbol{\theta}}$ and the true parameters vector $\boldsymbol{\theta}^*$ cannot be neglected. As a consequence, standard perturbative arguments that are the traditional basis for asymptotic normality fail. On the other hand, the Lasso estimator can be precisely characterized in the regime in which both $n$ and $p$ are large and $n/p$ is of order one. This characterization was first obtained in the case of Gaussian designs with i.i.d. covariates: here we generalize it to Gaussian correlated designs with non-singular covariance structure. This is expressed in terms of a simpler ``fixed-design'' model. We establish non-asymptotic bounds on the distance between the distribution of various quantities in the two models, which hold uniformly over signals $\boldsymbol{\theta}^*$ in a suitable sparsity class and over values of the regularization parameter. As an application, we study the distribution of the debiased Lasso and show that a degrees-of-freedom correction is necessary for computing valid confidence intervals.
相關內容
Quantifying the difference between two probability density functions, $p$ and $q$, using available data, is a fundamental problem in Statistics and Machine Learning. A usual approach for addressing this problem is the likelihood-ratio estimation (LRE) between $p$ and $q$, which -- to our best knowledge -- has been investigated mainly for the offline case. This paper contributes by introducing a new framework for online non-parametric LRE (OLRE) for the setting where pairs of iid observations $(x_t \sim p, x'_t \sim q)$ are observed over time. The non-parametric nature of our approach has the advantage of being agnostic to the forms of $p$ and $q$. Moreover, we capitalize on the recent advances in Kernel Methods and functional minimization to develop an estimator that can be efficiently updated online. We provide theoretical guarantees for the performance of the OLRE method along with empirical validation in synthetic experiments.
The use of the non-parametric Restricted Mean Survival Time endpoint (RMST) has grown in popularity as trialists look to analyse time-to-event outcomes without the restrictions of the proportional hazards assumption. In this paper, we evaluate the power and type I error rate of the parametric and non-parametric RMST estimators when treatment effect is explained by multiple covariates, including an interaction term. Utilising the RMST estimator in this way allows the combined treatment effect to be summarised as a one-dimensional estimator, which is evaluated using a one-sided hypothesis Z-test. The estimators are either fully specified or misspecified, both in terms of unaccounted covariates or misspecified knot points (where trials exhibit crossing survival curves). A placebo-controlled trial of Gamma interferon is used as a motivating example to simulate associated survival times. When correctly specified, the parametric RMST estimator has the greatest power, regardless of the time of analysis. The misspecified RMST estimator generally performs similarly when covariates mirror those of the fitted case study dataset. However, as the magnitude of the unaccounted covariate increases, the associated power of the estimator decreases. In all cases, the non-parametric RMST estimator has the lowest power, and power remains very reliant on the time of analysis (with a later analysis time correlated with greater power).
In prediction settings where data are collected over time, it is often of interest to understand both the importance of variables for predicting the response at each time point and the importance summarized over the time series. Building on recent advances in estimation and inference for variable importance measures, we define summaries of variable importance trajectories. These measures can be estimated and the same approaches for inference can be applied regardless of the choice of the algorithm(s) used to estimate the prediction function. We propose a nonparametric efficient estimation and inference procedure as well as a null hypothesis testing procedure that are valid even when complex machine learning tools are used for prediction. Through simulations, we demonstrate that our proposed procedures have good operating characteristics, and we illustrate their use by investigating the longitudinal importance of risk factors for suicide attempt.
In this paper we propose a definition of the distributional Riemann curvature tensor in dimension $N\geq 2$ if the underlying metric tensor $g$ defined on a triangulation $\mathcal{T}$ possesses only single-valued tangential-tangential components on codimension 1 simplices. We analyze the convergence of the curvature approximation in the $H^{-2}$-norm if a sequence of interpolants $g_h$ of polynomial order $k\geq 0$ of a smooth metric $g$ is given. We show that for dimension $N=2$ convergence rates of order $\mathcal{O}(h^{k+1})$ are obtained. For $N\geq 3$ convergence holds only in the case $k\geq 1$. Numerical examples demonstrate that our theoretical results are sharp. By choosing appropriate test functions we show that the distributional Gauss and scalar curvature in 2D respectively any dimension are obtained. Further, a first definition of the distributional Ricci curvature tensor in arbitrary dimension is derived, for which our analysis is applicable.
We consider the estimation of the cumulative hazard function, and equivalently the distribution function, with censored data under a setup that preserves the privacy of the survival database. This is done through a $\alpha$-locally differentially private mechanism for the failure indicators and by proposing a non-parametric kernel estimator for the cumulative hazard function that remains consistent under the privatization. Under mild conditions, we also prove lowers bounds for the minimax rates of convergence and show that estimator is minimax optimal under a well-chosen bandwidth.
We propose to approximate a (possibly discontinuous) multivariate function f (x) on a compact set by the partial minimizer arg miny p(x, y) of an appropriate polynomial p whose construction can be cast in a univariate sum of squares (SOS) framework, resulting in a highly structured convex semidefinite program. In a number of non-trivial cases (e.g. when f is a piecewise polynomial) we prove that the approximation is exact with a low-degree polynomial p. Our approach has three distinguishing features: (i) It is mesh-free and does not require the knowledge of the discontinuity locations. (ii) It is model-free in the sense that we only assume that the function to be approximated is available through samples (point evaluations). (iii) The size of the semidefinite program is independent of the ambient dimension and depends linearly on the number of samples. We also analyze the sample complexity of the approach, proving a generalization error bound in a probabilistic setting. This allows for a comparison with machine learning approaches.
We propose a method to modify a polygonal mesh in order to fit the zero-isoline of a level set function by extending a standard body-fitted strategy to a tessellation with arbitrarily-shaped elements. The novel level set-fitted approach, in combination with a Discontinuous Galerkin finite element approximation, provides an ideal setting to model physical problems characterized by embedded or evolving complex geometries, since it allows skipping any mesh post-processing in terms of grid quality. The proposed methodology is firstly assessed on the linear elasticity equation, by verifying the approximation capability of the level set-fitted approach when dealing with configurations with heterogeneous material properties. Successively, we combine the level set-fitted methodology with a minimum compliance topology optimization technique, in order to deliver optimized layouts exhibiting crisp boundaries and reliable mechanical performances. An extensive numerical test campaign confirms the effectiveness of the proposed method.
Austrin showed that the approximation ratio $\beta\approx 0.94016567$ obtained by the MAX 2-SAT approximation algorithm of Lewin, Livnat and Zwick (LLZ) is optimal modulo the Unique Games Conjecture (UGC) and modulo a Simplicity Conjecture that states that the worst performance of the algorithm is obtained on so called simple configurations. We prove Austrin's conjecture, thereby showing the optimality of the LLZ approximation algorithm, relying only on the Unique Games Conjecture. Our proof uses a combination of analytic and computational tools. We also present new approximation algorithms for two restrictions of the MAX 2-SAT problem. For MAX HORN-$\{1,2\}$-SAT, i.e., MAX CSP$(\{x\lor y,\bar{x}\lor y,x,\bar{x}\})$, in which clauses are not allowed to contain two negated literals, we obtain an approximation ratio of $0.94615981$. For MAX CSP$(\{x\lor y,x,\bar{x}\})$, i.e., when 2-clauses are not allowed to contain negated literals, we obtain an approximation ratio of $0.95397990$. By adapting Austrin's and our arguments for the MAX 2-SAT problem we show that these two approximation ratios are also tight, modulo only the UGC conjecture. This completes a full characterization of the approximability of the MAX 2-SAT problem and its restrictions.
Challenges to reproducibility and replicability have gained widespread attention over the past decade, driven by a number of large replication projects with lukewarm success rates. A nascent work has emerged developing algorithms to estimate, or predict, the replicability of published findings. The current study explores ways in which AI-enabled signals of confidence in research might be integrated into literature search. We interview 17 PhD researchers about their current processes for literature search and ask them to provide feedback on a prototype replicability estimation tool. Our findings suggest that information about replicability can support researchers throughout literature review and research design processes. However, explainability and interpretability of system outputs is critical, and potential drawbacks of AI-enabled confidence assessment need to be further studied before such tools could be widely accepted and deployed. We discuss implications for the design of technological tools to support scholarly activities and advance reproducibility and replicability.
Describing the equality conditions of the Alexandrov--Fenchel inequality has been a major open problem for decades. We prove that in the case of convex polytopes, this description is not in the polynomial hierarchy unless the polynomial hierarchy collapses to a finite level. This is the first hardness result for the problem, and is a complexity counterpart of the recent result by Shenfeld and van Handel (arXiv:archive/201104059), which gave a geometric characterization of the equality conditions. The proof involves Stanley's order polytopes and employs poset theoretic technology.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191