Ensemble methods such as bagging and random forests are ubiquitous in various fields, from finance to genomics. Despite their prevalence, the question of the efficient tuning of ensemble parameters has received relatively little attention. This paper introduces a cross-validation method, ECV (Extrapolated Cross-Validation), for tuning the ensemble and subsample sizes in randomized ensembles. Our method builds on two primary ingredients: initial estimators for small ensemble sizes using out-of-bag errors and a novel risk extrapolation technique that leverages the structure of prediction risk decomposition. By establishing uniform consistency of our risk extrapolation technique over ensemble and subsample sizes, we show that ECV yields $\delta$-optimal (with respect to the oracle-tuned risk) ensembles for squared prediction risk. Our theory accommodates general ensemble predictors, only requires mild moment assumptions, and allows for high-dimensional regimes where the feature dimension grows with the sample size. As a practical case study, we employ ECV to predict surface protein abundances from gene expressions in single-cell multiomics using random forests. In comparison to sample-split cross-validation and $K$-fold cross-validation, ECV achieves higher accuracy avoiding sample splitting. At the same time, its computational cost is considerably lower owing to the use of the risk extrapolation technique. Additional numerical results validate the finite-sample accuracy of ECV for several common ensemble predictors under a computational constraint on the maximum ensemble size.
相關內容
Block Principal Component Analysis (Block PCA) of a data matrix A, where loadings Z are determined by maximization of AZ 2 over unit norm orthogonal loadings, is difficult to use for the design of sparse PCA by 1 regularization, due to the difficulty of taking care of both the orthogonality constraint on loadings and the non differentiable 1 penalty. Our objective in this paper is to relax the orthogonality constraint on loadings by introducing new objective functions expvar(Y) which measure the part of the variance of the data matrix A explained by correlated components Y = AZ. So we propose first a comprehensive study of mathematical and numerical properties of expvar(Y) for two existing definitions Zou et al. [2006], Shen and Huang [2008] and four new definitions. Then we show that only two of these explained variance are fit to use as objective function in block PCA formulations for A rid of orthogonality constraints.
We study the problem of sequentially predicting properties of a probabilistic model and its next outcome over an infinite horizon, with the goal of ensuring that the predictions incur only finitely many errors with probability 1. We introduce a general framework that models such prediction problems, provide general characterizations for the existence of successful prediction rules, and demonstrate the application of these characterizations through several concrete problem setups, including hypothesis testing, online learning, and risk domination. In particular, our characterizations allow us to recover the findings of Dembo and Peres (1994) with simple and elementary proofs, and offer a partial resolution to an open problem posed therein.
By abstracting over well-known properties of De Bruijn's representation with nameless dummies, we design a new theory of syntax with variable binding and capture-avoiding substitution. We propose it as a simpler alternative to Fiore, Plotkin, and Turi's approach, with which we establish a strong formal link. We also show that our theory easily incorporates simple types and equations between terms.
The goal of anomaly detection is to identify observations that are generated by a distribution that differs from the reference distribution that qualifies normal behavior. When examining a time series, the reference distribution may evolve over time. The anomaly detector must therefore be able to adapt to such changes. In the online context, it is particularly difficult to adapt to abrupt and unpredictable changes. Our solution to this problem is based on the detection of breakpoints in order to adapt in real time to the new reference behavior of the series and to increase the accuracy of the anomaly detection. This solution also provides a control of the False Discovery Rate by extending methods developed for stationary series.
We present a subspace method to solve large-scale trace ratio problems. This method is matrix-free, only needing the action of the two matrices in the trace ratio. At each iteration, a smaller trace ratio problem is addressed in the search subspace. Additionally, our algorithm is endowed with a restarting strategy, that ensures the monotonicity of the trace ratio value throughout the iterations. We also investigate the behavior of the approximate solution from a theoretical viewpoint, extending existing results on Ritz values and vectors, as the angle between the search subspace and the exact solution to the trace ratio approaches zero. In the context of multigroup classification, numerical experiments show that the new subspace method tends to be more efficient than iterative approaches that need a (partial) eigenvalue decomposition in every step.
This paper introduces a novel concept of interval probability measures that enables the representation of imprecise probabilities, or uncertainty, in a natural and coherent manner. Within an algebra of sets, we introduce a notion of weak complementation denoted as $\psi$. The interval probability measure of an event $H$ is defined with respect to the set of indecisive eventualities $(\psi(H))^c$, which is included in the standard complement $H^c$. We characterize a broad class of interval probability measures and define their properties. Additionally, we establish an updating rule with respect to $H$, incorporating concepts of statistical independence and dependence. The interval distribution of a random variable is formulated, and a corresponding definition of stochastic dominance between two random variables is introduced. As a byproduct, a formal solution to the century-old Keynes-Ramsey controversy is presented.
We develop and implement methods for determining whether relaxing sparsity con- straints on portfolios improves the investment opportunity set for risk-averse investors. We formulate a new estimation procedure for sparse second-order stochastic spanning based on a greedy algorithm and Linear Programming. We show the optimal recovery of the sparse solution asymptotically whether spanning holds or not. From large equity datasets, we estimate the expected utility loss due to possible under-diversification, and find that there is no benefit from expanding a sparse opportunity set beyond 45 assets. The optimal sparse portfolio invests in 10 industry sectors and cuts tail risk when compared to a sparse mean-variance portfolio. On a rolling-window basis, the number of assets shrinks to 25 assets in crisis periods, while standard factor models cannot explain the performance of the sparse portfolios.
This manuscript bridges the divide between causal inference and spatial statistics, presenting novel insights for causal inference in spatial data analysis, and establishing how tools from spatial statistics can be used to draw causal inferences. We introduce spatial causal graphs to highlight that spatial confounding and interference can be entangled, in that investigating the presence of one can lead to wrongful conclusions in the presence of the other. Moreover, we show that spatial dependence in the exposure variable can render standard analyses invalid, which can lead to erroneous conclusions. To remedy these issues, we propose a Bayesian parametric approach based on tools commonly-used in spatial statistics. This approach simultaneously accounts for interference and mitigates bias resulting from local and neighborhood unmeasured spatial confounding. From a Bayesian perspective, we show that incorporating an exposure model is necessary, and we theoretically prove that all model parameters are identifiable, even in the presence of unmeasured confounding. To illustrate the approach's effectiveness, we provide results from a simulation study and a case study involving the impact of sulfur dioxide emissions from power plants on cardiovascular mortality.
A rectangulation is a decomposition of a rectangle into finitely many rectangles. Via natural equivalence relations, rectangulations can be seen as combinatorial objects with a rich structure, with links to lattice congruences, flip graphs, polytopes, lattice paths, Hopf algebras, etc. In this paper, we first revisit the structure of the respective equivalence classes: weak rectangulations that preserve rectangle-segment adjacencies, and strong rectangulations that preserve rectangle-rectangle adjacencies. We thoroughly investigate posets defined by adjacency in rectangulations of both kinds, and unify and simplify known bijections between rectangulations and permutation classes. This yields a uniform treatment of mappings between permutations and rectangulations that unifies the results from earlier contributions, and emphasizes parallelism and differences between the weak and the strong cases. Then, we consider the special case of guillotine rectangulations, and prove that they can be characterized - under all known mappings between permutations and rectangulations - by avoidance of two mesh patterns that correspond to "windmills" in rectangulations. This yields new permutation classes in bijection with weak guillotine rectangulations, and the first known permutation class in bijection with strong guillotine rectangulations. Finally, we address enumerative issues and prove asymptotic bounds for several families of strong rectangulations.
Most of the existing Mendelian randomization (MR) methods are limited by the assumption of linear causality between exposure and outcome, and the development of new non-linear MR methods is highly desirable. We introduce two-stage prediction estimation and control function estimation from econometrics to MR and extend them to non-linear causality. We give conditions for parameter identification and theoretically prove the consistency and asymptotic normality of the estimates. We compare the two methods theoretically under both linear and non-linear causality. We also extend the control function estimation to a more flexible semi-parametric framework without detailed parametric specifications of causality. Extensive simulations numerically corroborate our theoretical results. Application to UK Biobank data reveals non-linear causal relationships between sleep duration and systolic/diastolic blood pressure.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191