2025 年的期刊文章 · 第 38 页

共 1491 篇,按发表时间排,新的在前,每页 40 篇。这一页是 2025 到 2025。

按期刊、按主题看 →

第 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 页

JMLR Other · 发表于 2025

A Random Matrix Approach to Low-Multilinear-Rank Tensor Approximation

Hugo Lebeau, Florent Chatelain, Romain Couillet

摘要

This work presents a comprehensive understanding of the estimation of a planted low-rank signal from a general spiked tensor model near the computational threshold. Relying on standard tools from the theory of large random matrices, we characterize the large-dimensional spectral behavior of the unfoldings of the data tensor and exhibit relevant signal-to-noise ratios governing the detectability of the principal directions of the signal. These results allow to accurately predict the reconstruction performance of truncated multilinear SVD (MLSVD) in the non-trivial regime. This is particularly important since it serves as an initialization of the higher-order orthogonal iteration (HOOI) scheme, whose convergence to the best low-multilinear-rank approximation depends entirely on its initialization. We give a sufficient condition for the convergence of HOOI and show that the number of iterations before convergence tends to $1$ in the large-dimensional limit.

Econometrica Other · 发表于 2025

A Quest for Knowledge

Christoph Carnehl, Johannes Schneider

摘要

Is more novel research always desirable? We develop a model in which knowledge shapes society's policies and guides the search for discoveries. Researchers select a question and how intensely to study it. The novelty of a question determines both the value and difficulty of discovering its answer. We show that the benefits of discoveries are nonmonotone in novelty. Knowledge expands endogenously step‐by‐step over time. Through a dynamic externality, moonshots—research on questions more novel than what is myopically optimal—can improve the evolution of knowledge. Moonshots induce research cycles in which subsequent researchers connect the moonshot to previous knowledge.

JMLR Optimization & Learning Theory · 发表于 2025

A New Random Reshuffling Method for Nonsmooth Nonconvex Finite-sum Optimization

Junwen Qiu, Xiao Li, Andre Milzarek

摘要

Random reshuffling techniques are prevalent in large-scale applications, such as training neural networks. While the convergence and acceleration effects of random reshuffling-type methods are fairly well understood in the smooth setting, much less studies seem available in the nonsmooth case. In this work, we design a new normal map-based proximal random reshuffling (norm-PRR) method for nonsmooth nonconvex finite-sum problems. We show that norm-PRR achieves the iteration complexity ${\cal O}(n^{-1/3}T^{-2/3})$ where $n$ denotes the number of component functions $f(\cdot,i)$ and $T$ counts the total number of iterations. This improves the currently known complexity bounds for this class of problems by a factor of $n^{-1/3}$ in terms of the number of gradient evaluations. Additionally, we prove that norm-PRR converges linearly under the (global) Polyak-Łojasiewicz condition and in the interpolation setting. We further complement these non-asymptotic results and provide an in-depth analysis of the asymptotic properties of norm-PRR. Specifically, under the (local) Kurdyka-Łojasiewicz inequality, the whole sequence of iterates generated by norm-PRR is shown to converge to a single stationary point. Moreover, we derive last-iterate convergence rates that can match those in the smooth, strongly convex setting. Finally, numerical experiments are performed on nonconvex classification tasks to illustrate the efficiency of the proposed approach.

JMLR Other · 发表于 2025

A Hybrid Weighted Nearest Neighbour Classifier for Semi-Supervised Learning

Stephen M. S. Lee, Mehdi Soleymani

摘要

We propose a novel hybrid procedure for constructing a randomly weighted nearest neighbour classifier for semi-supervised learning. The procedure first uses the labelled learning set to predict a probability distribution of class labels for the unlabelled learning set. This turns the unlabelled set into a pseudo-labelled set, on which a sequentially weighted nearest neighbour classifier can be trained. The vote proportions calculated by this sequentially weighted nearest neighbour classifier and the standard weighted nearest neighbour classifier trained on the labelled set alone are then linearly combined to build a hybrid classifier. Our theory shows that, given a sufficiently large set of unlabelled data, the hybrid classifier has an optimal regret converging at a faster rate than that of the optimally weighted nearest neighbour classifier and hence of the optimal bagged or k-nearest neighbour classifier. We also show that the hybrid classifier can be revised by a dislabelling strategy to achieve the fastest possible rate of regret irrespective of the size of the unlabelled set, which may even be empty. Simulation studies and real data examples are presented to support our theoretical findings and illustrate the empirical performance of the hybrid classifiers constructed using uniform weights. We also explore the effects of pseudo-labelling by hypothesized class probabilities as a supplement to our main findings.

JMLR Optimization & Learning Theory · 发表于 2025

A Decentralized Proximal Gradient Tracking Algorithm for Composite Optimization on Riemannian Manifolds

Lei Wang, Le Bao, Xin Liu

摘要

This paper focuses on minimizing a smooth function combined with a nonsmooth regularization term on a compact Riemannian submanifold embedded in the Euclidean space under a decentralized setting. Typically, there are two types of approaches at present for tackling such composite optimization problems. The first, subgradient-based approaches, rely on subgradient information of the objective function to update variables, achieving an iteration complexity of $O(\epsilon^{-4}\log^2(\epsilon^{-2}))$. The second, smoothing approaches, involve constructing a smooth approximation of the nonsmooth regularization term, resulting in an iteration complexity of $O(\epsilon^{-4})$. This paper proposes a proximal gradient type algorithm that fully exploits the composite structure. The global convergence to a stationary point is established with a significantly improved iteration complexity of $O(\epsilon^{-2})$. To validate the effectiveness and efficiency of our proposed method, we present numerical results from real-world applications, showcasing its superior performance compared to existing approaches.

JMLR Other · 发表于 2025

A Comparative Evaluation of Quantification Methods

Tobias Schumacher, Markus Strohmaier, Florian Lemmerich

Econometrica Other · 发表于 2025

A Comment on: “Monotone Comparative Statics”

Rabah Amir, David Rietzke

摘要

Milgrom and Shannon (1994) provide necessary and sufficient conditions on parameterized optimization problems for their solution sets to be globally monotone in the parameter. We establish that their conditions may be significantly relaxed when focusing on discrete, binary comparisons between solution sets. Such binary comparisons are ubiquitous in economics and may involve comparing the same decision maker across two distinct regimes or two distinct decision makers with related objectives (e.g., a monopolist firm versus a social planner). While the single‐crossing property remains prominent in the theory, quasisupermodularity of the objective functions of interest is not needed. Our approach relies upon a novel method of embedding a new optimization problem with a quasisupermodular objective function “between” the two original problems of interest. In smooth problems, sufficient conditions for our new assumptions may be verified by elementary differential comparisons, making them well suited for applied work. We illustrate the relevance of this novel approach with several economic applications.

Econometrica Causal Inference Machine Learning · 发表于 2025

A Comment on: “Fisher–Schultz Lecture: Generic Machine Learning Inference on Heterogeneous Treatment Effects in Randomized Experiments, With an Application to Immunization in India” by Victor Chernozhukov, Mert Demirer, Esther Duflo, and Iván Fernández‐Val

Kosuke Imai, Michael Lingzhi Li

摘要

We examine the split‐sample robust inference (SSRI) methodology introduced by Chernozhukov, Demirer, Duflo, and Fernandez‐Val for quantifying uncertainty in heterogeneous treatment effect estimates produced by machine learning (ML) models. Although SSRI properly accounts for the additional variability due to sample splitting, its computational cost becomes prohibitive with complex ML models. We propose an alternative approach based on randomization inference (RI) that preserves the broad applicability of SSRI while eliminating the need for repeated sample splitting. Leveraging cross‐fitting and design‐based inference, the RI procedure yields valid confidence intervals with substantially reduced computational burden. Simulation studies demonstrate that the RI method preserves the statistical efficiency of SSRI while scaling to much larger applications and more complex settings.

Econometrica Spatial & Time Series · 发表于 2025

A Comment on: “Autoregressive Conditional Duration: A New Model for Irregularly Spaced Transaction Data”

Giuseppe Cavaliere, Thomas Mikosch, Anders Rahbek, Frederik Vilandt

摘要

Based on the GARCH literature, Engle and Russell (1998) established consistency and asymptotic normality of the QMLE for the autoregressive conditional duration (ACD) model, assuming strict stationarity and ergodicity of the durations. Using novel arguments based on renewal process theory, we show that their results hold under the stronger requirement that durations have finite expectation. However, we demonstrate that this is not always the case under the assumption of stationary and ergodic durations. Specifically, we provide a counterexample where the MLE is asymptotically mixed normal and converges at a rate significantly slower than usual. The main difference between ACD and GARCH asymptotics is that the former must account for the number of durations in a given time span being random. As a by‐product, we present a new lemma which can be applied to analyze asymptotic properties of extremum estimators when the number of observations is random.

JMLR Other · 发表于 2025

(De)-regularized Maximum Mean Discrepancy Gradient Flow

Zonghao Chen, Aratrika Mustafi, Pierre Glaser, Anna Korba, Arthur Gretton, Bharath K. Sriperumbudur

摘要

We introduce a (de)-regularization of the Maximum Mean Discrepancy (DrMMD) and its Wasserstein gradient flow. Existing gradient flows that transport samples from source distribution to target distribution with only target samples, either lack tractable numerical implementation ($f$-divergence flows) or require strong assumptions and modifications, such as noise injection, to ensure convergence (Maximum Mean Discrepancy flows). In contrast, DrMMD flow can simultaneously (i) guarantee near-global convergence for a broad class of targets in both continuous and discrete time, and (ii) be implemented in closed form using only samples. The former is achieved by leveraging the connection between the DrMMD and the $\chi^2$-divergence, while the latter comes by treating DrMMD as MMD with a de-regularized kernel. Our numerical scheme employs an adaptive de-regularization schedule throughout the flow to optimally balance the trade-off between discretization errors and deviations from the $\chi^2$ regime. The potential application of the DrMMD flow is demonstrated across several numerical experiments, including a large-scale setting of training student/teacher networks.

JMLR Other · 发表于 2025

"What is Different Between These Datasets?" A Framework for Explaining Data Distribution Shifts

Varun Babbar*, Zhicheng Guo*, Cynthia Rudin

摘要

The performance of machine learning models relies heavily on the quality of input data, yet real-world applications often face significant data-related challenges. A common issue arises when curating training data or deploying models: two datasets from the same domain may exhibit differing distributions. While many techniques exist for detecting such distribution shifts, there is a lack of comprehensive methods to explain these differences in a human-understandable way beyond opaque quantitative metrics. To bridge this gap, we propose a versatile framework of interpretable methods for comparing datasets. Using a variety of case studies, we demonstrate the effectiveness of our approach across diverse data modalities—including tabular data, text data, images, time-series signals – in both low and high-dimensional settings. These methods complement existing techniques by providing actionable and interpretable insights to better understand and address distribution shifts.

第 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 页