2026 年的期刊文章 · 第 37 页

共 1756 篇,按发表时间排,新的在前,每页 40 篇。这一页是 2026-01-05 到 2026。

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 39 40 41 42 43 44 页

JASA Econometrics & Economics Machine Learning Testing & Uncertainty · 发表于 2026-01-05

Online Auction Design Using Distribution-Free Uncertainty Quantification with Applications to E-Commerce

Jiale Han, Xiaowu Dai

摘要

Online auction is a cornerstone of e-commerce, and a key challenge is designing incentive-compatible mechanisms that maximize expected revenue. Existing approaches often assume known bidder value distributions and fixed sets of bidders and items, but these assumptions rarely hold in real-world settings where bidder values are unknown, and the number of future participants is uncertain. In this article, we introduce the Conformal Online Auction Design (COAD), a novel mechanism that maximizes revenue by quantifying uncertainty in bidder values without relying on known distributions. COAD incorporates both bidder and item features, using historical data to design an incentive-compatible mechanism for online auctions. Unlike traditional methods, COAD leverages distribution-free uncertainty quantification techniques and integrates machine learning methods, such as random forests, kernel methods, and deep neural networks, to predict bidder values while ensuring revenue guarantees. Moreover, COAD introduces bidder-specific reserve prices, based on the lower confidence bounds of bidder valuations, contrasting with the single reserve price commonly used in the literature. We demonstrate the practical effectiveness of COAD through an application to real-world eBay auction data. Theoretical results and extensive simulation studies further validate the properties of our approach. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work. *(abstract via PubMed)*

JASA Survival Analysis · 发表于 2026-01-05

Localizing Strictly Proper Scoring Rules

Ramon F. A. de Punder, Cees G. H. Diks, Roger J. A. Laeven, Dick J. C. van Dijk

摘要

When comparing predictive distributions, forecasters are typically not equally interested in all regions of the outcome space. To address the demand for focused forecast evaluation, we propose a procedure to transform strictly proper scoring rules into their localized counterparts while preserving the score divergence and strict propriety. This is accomplished by applying the original scoring rule to a censored distribution. Our procedure nests the censored likelihood score as a special case. Among a multitude of others, it also implies a class of censored kernel scores that offers a (possibly multivariate) alternative to the threshold weighted Continuously Ranked Probability Score (twCRPS), extending its local propriety to more general weight functions than single tail indicators. Within this localized framework, we obtain a generalization of the Neyman Pearson lemma, establishing the censored likelihood ratio test as uniformly most powerful. For other tests of localized equal predictive performance, results of Monte Carlo simulations and empirical applications to risk management, inflation and climate data consistently emphasize the excellent power properties of censoring versus other localization methods. *(abstract via OpenAlex)*

JASA Causal Inference · 发表于 2026-01-05

Identification and Multiply Robust Estimation of Causal Effects via Instrumental Variables from An Auxiliary Population

Wei Li, Jiapeng Liu, Peng Ding, Zhi Geng

摘要

Estimating causal effects in a target population with unmeasured confounders is challenging, especially when instrumental variables (IVs) are unavailable. However, IVs from auxiliary populations with similar problems can help infer causal effects in the target population. While the homogeneous conditional average treatment effect assumption has been widely used for effect transportability, it has not been explored in IV-based data fusion. We include it as a basic approach, though it may be biased when treatment effect heterogeneity exists. As an alternative approach, we introduce the equi-confounding assumption that the unmeasured confounding bias remains the same after adjusting for observed covariates, while allowing conditional average treatment effects to differ across populations. This allows us to identify the confounding bias in the auxiliary population and remove it from the treatment-outcome association in the target population to recover the causal effect. We develop multiply robust estimators under both approaches and demonstrate them through simulation studies and a real data application. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work. *(abstract via OpenAlex)*

JASA Bayesian Methods · 发表于 2026-01-05

Bayesian Image Analysis in Fourier Space

John Kornak, Karl Young, Eric Friedman, Konstantinos Bakas

摘要

Bayesian image analysis has been instrumental for over 40 years in addressing challenges such as image noise reduction, de-blurring, feature enhancement, and object detection. Despite its success, modeling spatial dependencies inherent to these problems often results in significant computational challenges. This work introduces the Bayesian Image Analysis in Fourier Space (BIFS) framework, which redefines conventional Bayesian modeling for continuous-valued images by transforming the problem into the Fourier domain. This transformation reduces the original high-dimensional dependent estimation problem into multiple low-dimensional, independent subproblems in Fourier space. The BIFS approach thereby simplifies computation while enabling flexible model specification, efficient formulation of isotropic priors, adaptability to diverse prior expectations, and invariance to changes in image resolution. BIFS thus offers a powerful and computationally efficient framework for a wide range of imaging applications. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work. *(abstract via PubMed)*

JASA Other · 发表于 2026-01-02

When Does Bottom-Up Beat Top-Down in Hierarchical Community Detection?

Maximilien Dreveton, Daichi Kuroda, Matthias Grossglauser, Patrick Thiran

摘要

Hierarchical community detection consists in finding a tree of communities where deeper levels of the hierarchy reveal finer-grained structures. There are two main classes of algorithms for this task. Divisive (top-down) algorithms recursively partition nodes into smaller communities until a stopping criterion indicates that no further splits are necessary. In contrast, agglomerative (bottom-up) algorithms first identify the smallest community structures and then repeatedly merge the communities by using a linkage method. In this work, we prove that a bottom-up algorithm recovers the hierarchy of a hierarchical stochastic block model (HSBM) when the average degree grows unbounded. We also establish the information-theoretic threshold for exact recovery at intermediate depths of the hierarchy and highlight its significance in understanding the limitations of top-down algorithms. Numerical experiments on both synthetic and real datasets demonstrate the superiority of bottom-up methods. In particular, a notable drawback of top-down algorithms is their tendency to produce dendrograms with inversions. These findings contribute to a better understanding of hierarchical clustering techniques and their applications in network analysis. *(abstract via OpenAlex)*

JASA Survival Analysis Nonparametric & Semiparametric Causal Inference · 发表于 2026-01-02

SPARCC: Semi-Parametric Robust Estimation in a Right-Censored Covariate Model

Seong-ho Lee, Brian D. Richardson, Yanyuan Ma, Karen S. Marder, Tanya P. Garcia

摘要

In Huntington disease research, a current goal is to understand how symptoms change prior to a clinical diagnosis. Statistically, achieving this goal entails modeling symptom severity as a function of the covariate 'time of diagnosis,' which is often heavily right-censored in observational studies. Existing estimators that handle right-censored covariates, such as the complete case estimator and maximum likelihood estimator, vary in their statistical efficiency and robustness to misspecifications of nuisance parameters (i.e., densities for the censored covariate and censoring variable). We propose a new "SPARCC" estimator (SemiPArametric Robust estimation in a right-Censored Covariate model) that is robust and efficient. When the nuisance parameters are modeled parametrically, the SPARCC estimator is doubly robust, i.e., consistent if at least one nuisance parameter is correctly specified, and semiparametric efficient if both are correctly specified. When the nuisance parameters are estimated via nonparametric or machine learning methods that converge sufficiently fast, the SPARCC estimator is consistent and semiparametric efficient. We show empirically that the proposed estimator, implemented in the R package sparcc, has its claimed properties, and we apply it to estimate Huntington disease symptom trajectories using data from the Enroll-HD study. *(abstract via PubMed)*

SMMR Longitudinal Data Latent Variables & Mixtures Survival Analysis · 发表于 2026-01-02

Joint model with latent disease age: Overcoming the need for reference time

Juliette Ortholand, Nicolas Gensollen, Stanley Durrleman, Sophie Tezenas Du Montcel

摘要

Heterogeneity of the progression of neurodegenerative diseases is one of the main challenges faced in developing therapies. Thanks to the increasing number of clinical databases, progression models have allowed a better understanding of this heterogeneity. Joint models have proven their effectiveness by combining longitudinal and survival data. Nevertheless, they require a reference time, which is ill-defined for neurodegenerative diseases, where biological underlying processes start before the first symptoms. In this work, we propose a joint non-linear mixed-effect model with a latent disease age, to overcome this need for a precise reference time. We used a longitudinal model with a latent disease age as a longitudinal sub-model. We associated it with a survival sub-model that estimates a Weibull distribution from the latent disease age. We validated our model on simulated data and benchmarked it with a state-of-the-art joint model on data from patients with Amyotrophic Lateral Sclerosis (ALS). Finally, we showed how the model could be used to describe ALS heterogeneity. Our model got significantly better results than the state-of-the-art joint model for absolute bias on ALS functional rating scale revised score (4.21(SD 4.41) versus 4.24(SD 4.14)( p -value = 1.4 × 10 − 17 )), and mean-cumulative-AUC for right-censored events on death (0.67(0.07) versus 0.61(0.09)( p -value = 1.7 × 10 − 03 )). To conclude, we propose a new model better suited in the context of unreliable reference time.

SMMR Survival Analysis · 发表于 2026-01-02

Functional varying-coefficient Cox model and its application

Fansheng Kong, Maozai Tian, Zhihao Wang, Man-lai Tang

摘要

When data become increasingly complex, desirable models are required to be more flexible for analyzing survival data. Building upon the existing functional Cox model, we introduce a novel functional varying-coefficient Cox model and the corresponding estimation algorithms are proposed in this article. The proposed model can simultaneously handle survival data with varying-coefficient covariates and functional covariates, thereby significantly enhancing the adaptability of survival models. The model performance is evaluated by simulation studies, and a real application using Alzheimer’s disease neuroimaging initiative (ADNI) data is used to illustrate the practicality of the proposed model.

JASA Other · 发表于 2026-01-02

Consistent Least Squares Estimation in Population-Size-Dependent Branching Processes

Peter Braunsteins, Sophie Hautphenne, Carmen Minuesa

摘要

We derive the first conditionally consistent estimators for a class of parametric Markov population models with logistic growth, which are suitable for modelling endangered populations in restricted habitats with a carrying capacity. We focus on discrete-time parametric population-size-dependent branching processes, for which we propose a new class of weighted least-squares estimators based on a single trajectory of population size counts. We establish the consistency and asymptotic normality of our estimators, conditional on non-extinction up to time n, as n→∞. Since Markov population models with a carrying capacity become extinct almost surely under general conditions, our proofs rely on arguments distinct from those in the existing literature.Our results are motivated by conservation biology, where endangered populations are often studied precisely because they are still alive, leading to an observation bias. Through simulated examples, we show that our conditionally consistent estimators generally reduce this bias for key quantities such as a habitat’s carrying capacity. We apply our methodology to estimate the carrying capacity of the Chatham Island black robin, a population reduced to a single breeding female in the 1970’s, which has since recovered but has yet to reach the island’s carrying capacity. *(abstract via OpenAlex)*

JASA Causal Inference · 发表于 2026-01-02

Causality-Oriented Robustness: Exploiting General Noise Interventions

Xinwei Shen, Peter Bühlmann, Armeen Taeb

摘要

Since distribution shifts are common in real-world applications, there is a pressing need to develop prediction models that are robust against such shifts. Existing frameworks, such as empirical risk minimization or distributionally robust optimization, either lack generalizability for unseen distributions or rely on postulated distance measures. Alternatively, causality offers a data-driven and structural perspective to robust predictions. However, the assumptions necessary for causal inference can be overly stringent, and the robustness offered by such causal models often lacks flexibility. In this article, we focus on causality-oriented robustness and propose Distributional Robustness via Invariant Gradients (DRIG), a method that exploits general noise interventions in training data for robust predictions against unseen interventions, and naturally interpolates between in-distribution prediction and causality. In a linear setting, we prove that DRIG yields predictions that are robust among a data-dependent class of distribution shifts. Furthermore, we show that our framework includes anchor regression as a special case, and that it yields prediction models that protect against more diverse perturbations. We establish finite-sample results and extend our approach to semi-supervised domain adaptation to further improve prediction performance. Finally, we empirically validate our methods on synthetic simulations and on single-cell and intensive health care datasets. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work. *(abstract via OpenAlex)*

JMLR Other · 发表于 2026,收录于 2026-09-10

torchsom: The Reference PyTorch Library for Self-Organizing Maps

Louis Berthier, Ahmed Shokry, Maxime Moreaud, Guillaume Ramelet, Eric Moulines

摘要

(verbatim, from arXiv preprint arxiv.org/abs/2510.11147 — the JMLR page itself could not be re-fetched to confirm word-for-word identity): "This paper introduces torchsom, an open-source Python library that provides a reference implementation of the Self-Organizing Map (SOM) in PyTorch. This package offers three main features: (i) dimensionality reduction, (ii) clustering, and (iii) friendly data visualization. It relies on a PyTorch backend, enabling (i) fast and efficient training of SOMs through GPU acceleration, and (ii) easy and scalable integration with the PyTorch ecosystem. torchsom also follows the scikit-learn API for ease of use and extensibility. The library is released under the Apache 2.0 license with 90% test coverage, and its source code and documentation are available at https://github.com/michelin/TorchSOM.

JMLR Other · 发表于 2026

torchgfn: A PyTorch GFlowNet Library

Joseph D. Viviano, Omar G. Younis, Sanghyeok Choi, Victor Schmidt, Yoshua Bengio, Salem Lahlou

摘要

The growing popularity of generative flow networks (GFlowNets or GFNs) among a range of researchers with diverse backgrounds and areas of expertise necessitates a library that facilitates the testing of new features (e.g., training losses and training policies) against standard benchmark implementations, or on a set of common environments. We present torchgfn, a PyTorch library that aims to address this need. Its core contribution is a modular and decoupled architecture which treats environments, neural network modules, and training objectives as interchangeable components. This provides users with a simple yet powerful API to facilitate rapid prototyping and novel research. Multiple examples are provided, replicating and unifying published results. The library is available on GitHub (https://github.com/GFNOrg/torchgfn) and on PyPI (https://pypi.org/project/torchgfn/).

JMLR Optimization & Learning Theory Machine Learning · 发表于 2026

skwdro: a library for Wasserstein distributionally robust machine learning

Florian Vincent, Waïss Azizian, Franck Iutzeler, Jérôme Malick

摘要

We present skwdro, a Python library for training robust machine learning models. The library is based on distributionally robust optimization using Wasserstein distances, popular in optimal transport and machine learnings. The goal of the library is to make the training of robust models easier for a wide audience by proposing a wrapper for PyTorch modules, enabling model loss' robustification with minimal code changes. It comes along with scikit-learn compatible estimators for some popular objectives. The core of the implementation relies on an entropic smoothing of the original robust objective, in order to ensure maximal model flexibility. The library is available at https://github.com/iutzeler/skwdro and the documentation at https://skwdro.readthedocs.io

JMLR Other · 发表于 2026

scikit-activeml: A Comprehensive and User-Friendly Active Learning Library

Marek Herde, Minh Tuan Pham, Daniel Kottke, Alexander Benz, Lukas Lührs, Pascal Mergard, Christoph Sandrock, Jiaying Cheng, Atal Roghman, Mehmet Müjde, Lukas Rauch, Bernhard Sick

摘要

scikit-activeml is a user-friendly open-source Python library for active learning on top of scikit-learn. Included are implementations of a large collection of query strategies, models, and visualization tools in pool- and stream-based active learning for classification or regression tasks with single or multiple annotators. The flexible design of the active learning cycle enables individual adaptations to a variety of learning scenarios. Our source code with comprehensive documentation is available at https://scikit-activeml.github.io.

JMLR Bayesian Methods Machine Learning Networks & Graphs · 发表于 2026

py/cuTAGI: An Open-Source Library for Tractable Approximate Gaussian Inference in Bayesian Neural Networks

Luong-Ha Nguyen, James-A. Goulet, Miquel Florensa-Montilla, Van-Dai Vuong

摘要

This paper introduces pyTAGI, a Python wrapper, and cuTAGI, its high-performance C++/CUDA backend, implementing Tractable Approximate Gaussian Inference (TAGI) for neural networks. TAGI treats all network quantities as Gaussian random variables and derives closed-form expressions for prior/posterior expected values, variances, and covariances, enabling analytic Bayesian learning without relying on gradient descent or backpropagation. The libraries mimic PyTorch's sequential interface, allowing users to define models by stacking layers in order and performing uncertainty-aware Bayesian inference. Beyond epistemic uncertainty, it also allows quantifying heteroscedastic aleatoric uncertainty. cuTAGI's custom CPU/GPU kernels and distributed-data-parallel support via NCCL/MPI deliver competitive runtimes, while pyTAGI's pip-installable frontend and MIT-licensed GitHub repo facilitate community adoption and extension. Version 0.2.1 already supports a comprehensive suite of layers and activations; future work will add eager execution, further kernel optimizations, attention mechanisms, and advanced covariance factorization. Together, py/cuTAGI offer an efficient, open-source foundation for the analytic treatment of Bayesian deep learning.

Econometrica Econometrics & Economics · 发表于 2026

Work Hours Mismatch

Marta Lachowska, Alexandre Mas, Raffaele Saggio, Stephen A. Woodbury

摘要

We apply a revealed preference approach to administrative data from Washington to show that constraints on work hours are widespread: Workers have limited discretion over hours at a given employer, and there is substantial mismatch of workers who prefer long hours to employers that provide short hours. Voluntary job transitions imply a ratio of the marginal rate of substitution of earnings for hours to the wage rate of 0.5–0.6 for prime‐age workers. The average absolute deviation between observed and optimal hours is about 15%, and low‐wage workers face particularly acute constraints on hours. On average, observed hours tend to be less than preferred levels, and workers would require a 12% higher wage with their current employer to be as well off as with an employer offering ideal hours. These findings suggest that hour constraints are an equilibrium feature of the labor market because long‐hour jobs are costly to employers.

Econometrica Econometrics & Economics · 发表于 2026,收录于 2026-09-10

Why Do Workers Dislike Inflation? Wage Erosion and Conflict Costs

Joao Guerreiro, Jonathon Hazell, Chen Lian, Christina Patterson

摘要

How costly is inflation to workers? Answers to this question have focused on the path of real wages during inflationary periods. We argue that workers must take costly actions (“conflict”) to have nominal wages catch up with inflation, meaning there are welfare costs even if real wages do not fall as inflation rises. We study a menu‐cost style model, where workers choose whether to engage in conflict with employers to secure a wage increase. We show that, following a rise in inflation, wage catch‐up resulting from more frequent conflict does not raise welfare. Instead, the impact of inflation on worker welfare is determined by what we call “wage erosion”—how inflation would lower real wages if workers' conflict decisions did not respond to inflation. As a result, using observed wage growth to measure worker welfare understates the costs of inflation. We conduct a survey showing that workers are willing to sacrifice around 1.75% of their wages to avoid conflict. Calibrating the model to survey data, we find that incorporating conflict significantly raises the costs of inflation for workers.

JMLR Machine Learning · 发表于 2026

Why "Classic" Transformers Are Shallow and A Depth-Enabling Technique

Yueyao Yu, Yin Zhang

摘要

Since its introduction in 2017, the Transformer has emerged as the leading neural network architecture, catalyzing revolutionary advancements in many AI disciplines. The key innovation in Transformer is a Self-Attention (SA) mechanism designed to capture contextual information. However, stacking up more layers of the same design has failed to produce trainable deeper Transformers. Thus far, various architectural modifications to the original design have been proposed to enable deeper depths for Transformer models, but a thorough understanding of this depth issue remains lacking. In this paper, we conduct a comprehensive investigation to substantiate the claim that the depth problem is caused by a phenomenon called token similarity escalation; that is, tokens grow increasingly alike after repeated applications of the SA mechanism. Our analysis reveals that, driven by the invariant leading eigenspace and large spectral gaps of attention matrices, token similarity provably escalates at a linear rate as the depth increases. This insight suggests a simple technique that surgically removes excessive token similarity without reducing the overall role of the SA mechanism, as is done by existing approaches. We perform a set of proof- of-concept, small-scale experiments to show the viability of the proposed depth-enabling technique.

JMLR Optimization & Learning Theory Networks & Graphs Machine Learning · 发表于 2026,收录于 2026-11-06

Weisfeiler-Lehman Meets Optimal Transport: Theory, Algorithms, and Implications

Samantha Chen, Sunhyuk Lim, Facundo Mémoli, Zhengchao Wan, Yusu Wang

摘要

We introduce the Weisfeiler–Lehman (WL) distance, a pseudometric defined on labeled measure Markov chains (LMMCs)—a general class of structures encompassing labeled graphs. This distance is computable in polynomial time and is closely aligned with the classical WL graph isomorphism test: it assigns a positive value if and only if the WL test distinguishes the two input graphs. The WL distance captures fine structural differences in the underlying LMMCs and is, consequently, more discriminative than the state-of-the-art Wasserstein WL graph kernel. The distance motivates a neural network architecture on LMMCs, which we call Markov chain neural networks (MCNNs), that is universal for continuous functions over LMMCs (hence all graphs) under the topology induced by the WL distance. This finding further extends to the analysis of graph neural networks (GNNs). Additionally, the WL distance admits an alternative characterization through Markovian couplings and Causal Optimal Transport (COT), offering a stochastic-process perspective that uncovers a surprising connection between the seemingly unrelated WL test and COT. Finally, we show that the WL distance is stable under a natural variant of the Gromov-Wasserstein (GW) distance that we propose. The GW distance is generally NP-hard to compute and hence the WL distance provides a polynomial-time computable lower bound for it.

Econometrica Econometrics & Economics · 发表于 2026

Walras–Bowley Lecture: Climate Policy in the Wide World

John Hassler, Per Krusell, Conny Olovsson

摘要

We construct a dynamic integrated assessment model of climate and the economy with very high geographic resolution. Migration is free within, but not allowed across, countries. The model parameterization uses a wealth of data, including the distribution of output, population, energy sources and use, and estimates of the local damages from climate change. It implies very large geographic dispersion in damages from warming. We conduct three kinds of policy experiments. In one, we note that a modest, uniform carbon tax limits global warming and damages around the world substantially. In a second experiment, we let the poorest countries not tax carbon, while the rest compensate by setting higher taxes; the efficiency losses are large. In a final experiment, we find that fast green technology growth alone is a poor substitute for carbon taxes, whether globally available or not.

JMLR Networks & Graphs · 发表于 2026

Viscosity Convergence Analysis for Deep Q-Networks

Qian Qi

摘要

Deep Q-Networks (DQNs) and related residual neural architectures are increasingly used for continuous-time reinforcement learning (CTRL), where optimal value functions solve fully nonlinear second-order Hamilton--Jacobi--Bellman (HJB) equations and may be non-smooth. In this regime, convergence should be analyzed in the viscosity-solution framework. We study a spatially-coupled monotone ResNet architecture whose one-step operator is a non-negative local aggregation, designed to satisfy monotonicity and stability at the operator level. Under a consistency assumption on the learned operator, we obtain operator-level convergence to the viscosity solution via the Barles--Souganidis framework. We then distinguish this idealized operator iteration from practical fitted value iteration with projection onto a neural function class, and discuss the resulting approximation gap. Empirically, on stochastic control benchmarks, the proposed architecture exhibits smaller overshoot near kinks and more stable numerical behavior than pointwise PINN-style baselines in our experiments.

JMLR Other · 发表于 2026

Vector-Valued Gaussian Processes for Approximating Divergence- or Rotation-free Vector Fields

Quoc Thong Le Gia, Ian Hugh Sloan, Holger Wendland

摘要

In this paper, we discuss vector-valued Gaussian processes for the approximation of divergence- or rotation-free functions. We establish the theory for such Gaussian processes, then link the theory to multivariate approximation theory, and finally give error estimates for the predictive mean in various situations.

JMLR Other · 发表于 2026

Vecchia-Inducing-Points Full-Scale Approximations for Gaussian Processes

Tim Gyger, Reinhard Furrer, Fabio Sigrist

摘要

Gaussian processes are flexible, probabilistic, non-parametric models widely used in machine learning and statistics. However, their scalability to large data sets is limited by computational constraints. To overcome these challenges, we propose Vecchia-inducing-points full-scale (VIF) approximations combining the strengths of global inducing points and local Vecchia approximations. Vecchia approximations excel in settings with low-dimensional inputs and moderately smooth covariance functions, while inducing point methods are better suited to high-dimensional inputs and smoother covariance functions. Our VIF approach bridges these two regimes by using an efficient correlation-based neighbor-finding strategy for the Vecchia approximation of the residual process, implemented via a modified cover tree algorithm. We further extend our framework to non-Gaussian likelihoods by introducing iterative methods that substantially reduce computational costs for training and prediction by several orders of magnitude compared to Cholesky-based computations when using a Laplace approximation. In particular, we propose and compare novel preconditioners and provide theoretical convergence results. Extensive numerical experiments on simulated and real-world data sets show that VIF approximations are both computationally efficient as well as more accurate and numerically stable than state-of-the-art alternatives. All methods are implemented in the open-source C++ library GPBoost with high-level Python and R interfaces.

SiM Causal Inference · 发表于 2026,收录于 2026-08-17

Using Negative Control Outcomes to Detect Selection Bias in Mendelian Randomization Studies

Apostolos Gkatzionis, George Davey Smith, Kate Tilling

摘要

Mendelian randomization is currently mainly implemented through the use of genetic variants as instrumental variables to inestigate the causal effect of an exposure on an outcome of interest. Mendelian randomization studies are robust to confounding bias and reverse causation, but they remain susceptible to selection bias; for example, this can happen if the exposure or outcome are associated with selection into the study sample. Negative controls are sometimes used to detect biases (typically due to confounding) in observational studies. Here, we focus specifically on Mendelian randomization analyses and discuss under what conditions a variable can be used as a negative control outcome to detect selection mechanisms that could bias Mendelian randomization estimates. We show that the main requirement is that the negative control outcome relates to confounders of the exposure and outcome. Counter-intuitively, the effect of the negative control on selection is of secondary concern; for example, a variable that does not affect selection can be a valid negative control for an outcome that does. We also investigate under what conditions age and sex can be used as negative control outcomes in Mendelian randomization analyses. In a real-data application, we investigate the pairwise causal relationships between 19 traits, utilizing data from the UK Biobank. Treating biological sex and natural hair color as negative control outcomes, we identify selection bias in analyses involving commonly used traits such as alcohol consumption, body mass index and educational attainment.

JMLR Machine Learning Bayesian Methods Latent Variables & Mixtures · 发表于 2026,收录于 2026-09-17

Unveiling the Statistical Foundations of Chain-of-Thought Prompting Methods

Xinyang Hu, Fengzhuo Zhang, Siyu Chen, Zhuoran Yang

摘要

Chain-of-Thought (CoT) prompting and its variants have gained popularity as effective methods for solving multi-step reasoning problems using pretrained large language models (LLMs). In this work, we analyze CoT prompting from a statistical estimation perspective, providing a comprehensive characterization of its sample complexity. To this end, we introduce a multi-step latent variable model that encapsulates the reasoning process, where the latent variable encodes the task information. Under this framework, we demonstrate that when the pretraining dataset is sufficiently large, the estimator formed by CoT prompting is equivalent to a Bayesian estimator. This estimator effectively solves the multi-step reasoning problem by aggregating a posterior distribution inferred from the demonstration examples in the prompt. Moreover, we prove that the statistical error of the CoT estimator can be decomposed into two main components: (i) a prompting error, which arises from inferring the true task using CoT prompts, and (ii) the statistical error of the pretrained LLM. We establish that, under appropriate assumptions, the prompting error decays exponentially to zero as the number of demonstrations increases. Additionally, we explicitly characterize the approximation and generalization errors of the pretrained LLM. Notably, we construct a transformer model that approximates the target distribution of the multi-step reasoning problem with an error that decreases exponentially in the number of transformer blocks. Our analysis extends to other variants of CoT, including Self-Consistent CoT, Tree-of-Thought, and Selection-Inference, offering a broad perspective on the efficacy of these methods. We also provide numerical experiments to validate the theoretical findings." (JMLR's own per-paper page was blocked; retrieved from the paper's arXiv mirror, arXiv:2408.14511.) *Publication/online date: date unknown (JMLR page inaccessible; listed in Volume 27, 2026)*

JMLR High-Dimensional · 发表于 2026

Unsupervised Feature Selection via Nonnegative Orthogonal Constrained Regularized Minimization

Yan Li, Defeng Sun, Liping Zhang

摘要

Unsupervised feature selection has drawn wide attention in the era of big data, since it serves as a fundamental technique for dimensionality reduction. However, many existing unsupervised feature selection models and solution methods are primarily designed for practical applications, and often lack rigorous theoretical support, such as convergence guarantees. In this paper, we first establish a novel unsupervised feature selection model based on regularized minimization with nonnegative orthogonality constraints, which has advantages of embedding feature selection into the nonnegative spectral clustering and preventing overfitting. To solve the proposed model, we develop an effective inexact augmented Lagrangian multiplier method, in which the subproblems are addressed using a proximal alternating minimization approach. We rigorously prove the algorithm's sequence converges to a stationary point of the model. Extensive numerical experiments on popular datasets demonstrate the stability and robustness of our method. Moreover, comparative results show that our method outperforms some existing state-of-the-art methods in terms of clustering evaluation metrics. The code is available at https://github.com/liyan-amss/NOCRM_code.

JMLR Bayesian Methods Optimization & Learning Theory · 发表于 2026

Underdamped Langevin MCMC with third order convergence

Maximilian Scott, Dáire O'Kane, Andraž Jelinčič, James Foster

摘要

In this paper, we propose a new numerical method for the underdamped Langevin diffusion (ULD) and present a non-asymptotic analysis of its sampling error in the 2-Wasserstein distance when the $d$-dimensional target distribution $p(x)\propto e^{-f(x)}$ is strongly log-concave and has varying degrees of smoothness. Precisely, under the assumptions that the gradient and Hessian of $f$ are Lipschitz continuous, our algorithm achieves a 2-Wasserstein error of $\varepsilon$ in $\mathcal{O}\big(\sqrt{d}/\varepsilon\big)$ and $\mathcal{O}\big(\sqrt{d}/\sqrt{\varepsilon}\big)$ steps respectively. Therefore, our algorithm has a similar complexity as other popular Langevin MCMC algorithms under matching assumptions. However, if we additionally assume that the third derivative of $f$ is Lipschitz continuous, then our algorithm achieves a 2-Wasserstein error of $\varepsilon$ in $\mathcal{O}\big(\sqrt{d}/\varepsilon^{\frac{1}{3}}\big)$ steps. To the best of our knowledge, this is the first gradient-only method for ULD with third order convergence. To support our theory, we perform Bayesian logistic regression across a range of real-world datasets, where our algorithm achieves competitive performance compared to an existing underdamped Langevin MCMC algorithm and the popular No U-Turn Sampler (NUTS).

JMLR Machine Learning · 发表于 2026

UQLM: A Python Package for Uncertainty Quantification in Large Language Models

Dylan Bouchard, Mohit Singh Chauhan, David Skarbrevik, Ho-Kyeong Ra, Viren Bajaj, Zeya Ahmad

摘要

Hallucinations, defined as instances where Large Language Models (LLMs) generate false or misleading content, pose a significant challenge that impacts the safety and trust of downstream applications. We introduce UQLM, a Python package for LLM hallucination detection using state-of-the-art uncertainty quantification (UQ) techniques. This toolkit offers a suite of UQ-based scorers that compute response-level confidence scores ranging from 0 to 1. This library provides an off-the-shelf solution for UQ-based hallucination detection that can be easily integrated to enhance the reliability of LLM outputs.

JMLR Networks & Graphs · 发表于 2026

Two-way Node Popularity Model for Directed and Bipartite Networks

Bing-Yi Jing, Ting Li, Jiangzhou Wang, Ya Wang

摘要

There has been increasing research attention on community detection in directed and bipartite networks. However, these studies often fail to consider the popularity of nodes in different communities, which is a common phenomenon in real-world networks. To address this issue, we propose a new probabilistic framework called the Two-Way Node Popularity Model (TNPM). The TNPM also accommodates edges from different distributions within a general sub-Gaussian family. We introduce the Delete-One-Method (DOM) for model fitting and community structure identification, and provide a comprehensive theoretical analysis with novel technical skills dealing with sub-Gaussian generalization. Additionally, we propose the Two-Stage Divided Cosine Algorithm (TSDC) to handle large-scale networks more efficiently. Our proposed methods offer multi-folded advantages in terms of estimation accuracy and computational efficiency, as demonstrated through extensive numerical studies. We apply our methods to two real-world applications, uncovering interesting findings.

JMLR Machine Learning · 发表于 2026

Transformers Can Overcome the Curse of Dimensionality: A Theoretical Study from an Approximation Perspective

Yuling Jiao, Yanming Lai, Yang Wang, Bokai Yan

摘要

The Transformer model is widely used in various application areas of machine learning, such as natural language processing. This paper investigates the approximation of the Hölder continuous function class $\mathcal{H}_{Q}^{\beta}\left([0,1]^{d\times n},\mathbb{R}^{d\times n}\right)$ by Transformers and constructs several Transformers that can overcome the curse of dimensionality. These Transformers consist of one self-attention layer with one head and the softmax function as the activation function, along with several feedforward layers. For example, to achieve an approximation accuracy of $\epsilon$, if the activation functions of the feedforward layers in the Transformer are ReLU and floor, only $\mathcal{O}\left(\log\frac{1}{\epsilon}\right)$ layers of feedforward layers are needed, with widths of these layers not exceeding $\mathcal{O}\left(\frac{1}{\epsilon^{2/\beta}}\log\frac{1}{\epsilon}\right)$. If other activation functions are allowed in the feedforward layers, the width of the feedforward layers can be further reduced to a constant. These results demonstrate that Transformers have a strong expressive capability. The construction in this paper is based on the Kolmogorov-Arnold Superposition Theorem and does not require the concept of contextual mapping, hence our proof is more intuitively clear compared to previous Transformer approximation works. Additionally, the translation technique proposed in this paper helps to apply the previous approximation results of feedforward neural networks to Transformer research.

JMLR Meta-Analysis & Data Integration · 发表于 2026

Transfer Learning via Regularized Random-effects Linear Discriminant Analysis

Hongzhe Zhang, Arnab Auddy, Hongzhe Li

摘要

Linear discriminant analysis is a widely used method for classification. However, the high dimensionality of predictors combined with small sample sizes often results in large classification errors. To address this challenge, it is crucial to leverage data from related source models to enhance the classification performance of a target model. This paper proposes a transfer learning approach via regularized random-effects linear discriminant analysis, where the discriminant direction is estimated as a weighted combination of ridge estimates obtained from both the target and source models. Multiple strategies for determining these weights are introduced and evaluated, including one that minimizes the estimation risk of the discriminant vector and another that minimizes the classification error. Utilizing results from random matrix theory, we explicitly derive the asymptotic values of these weights and the associated classification error rates in the high-dimensional setting, where the aspect ratio $\gamma := p/n$ as $p, n\rightarrow \infty$, with $p$ representing the predictor dimension and $n$ the sample size. Extensive numerical studies, including simulations, the analysis of the proteomics-based cardiovascular disease risk classification and the lipid traits classification problem with genotype data, demonstrate the effectiveness of the proposed approach.

JMLR Testing & Uncertainty · 发表于 2026

Transfer Conformal Predictive Inference for Regression

Ce Zhang, Ting Li, Jinhan Xie, Linglong Kong, Bei Jiang

摘要

Conformal prediction, a powerful framework for constructing prediction intervals for response variables using any regression function estimators, often faces the challenge of producing overly broad intervals with limited target data. In this paper, we study the transfer learning problem in conformal prediction, aiming to improve the precision of the prediction interval of the target data with insufficient data by leveraging related auxiliary source datasets. Allowing for the potential non-exchangeability between source and target datasets, we propose two transfer conformal prediction algorithms designed for scenarios where knowledge of informative source data is either present or absent. Our approach uses conditional Kullback-Leibler divergence to effectively identify relevant source datasets for transfer. A comprehensive theoretical analysis of the non-asymptotic properties of the proposed algorithms is provided, including lower and upper bounds, and the prediction interval width. These results illustrate the potential to achieve more efficient, narrower intervals without compromising coverage accuracy. Empirical results from extensive simulations and real-world data confirm the efficacy of our methods, demonstrating significant improvements in prediction interval precision by leveraging source data, achieving narrower intervals while maintaining desired coverage levels.

Econometrica Other · 发表于 2026

Training Specificity and Occupational Mobility: Evidence From German Apprenticeships

Dita Eckardt

摘要

Apprenticeships play a key role in enabling successful school‐to‐work transitions in many countries, but in the presence of imperfect information, the specificity of this type of training may entail important costs for those working outside their training fields. I study this issue in one of the most prominent training settings, the German apprenticeship system. Using administrative data and a broad occupational classification, I find that 40% of individuals work in occupations different from their training. I estimate the cost of mismatch using vacancy instruments and extend methodological approaches in high‐dimensional selection settings. Lacking training in one's occupation entails an average wage penalty of 14%, the equivalent of two years of work experience. The penalty increases with the task distance between training and occupation. My findings suggest that retraining is crucial to mitigate the adverse consequences from imperfect information in specialized training settings.

Econometrica Econometrics & Economics · 发表于 2026

Trade and Domestic Distortions: The Case of Informality

Rafael Dix-Carneiro, Pinelopi Goldberg, Costas Meghir, Gabriel Ulyssea

摘要

We examine the effects of international trade in the presence of a set of domestic distortions giving rise to informality, a prevalent phenomenon in developing countries. In our quantitative model, the informal sector arises from burdensome taxes and regulations that are imperfectly enforced by the government. In equilibrium, smaller, less productive firms face fewer distortions than larger, more productive ones, potentially leading to substantial misallocation. We show that in settings with a large informal sector, the gains from trade are significantly amplified, as reductions in trade barriers imply a reallocation of resources from initially less distorted to more distorted firms. We confirm findings from earlier reduced‐form studies that the informal sector mitigates the impact of negative labor demand shocks on unemployment. Nonetheless, the informal sector can exacerbate the adverse real income effects of economic downturns, amplifying misallocation. Last, our research sheds light on the relationship between trade openness and cross‐firm wage inequality.

JMLR Other · 发表于 2026

Towards Convexity in Anomaly Detection: A New Formulation of SSLM with Unique Optimal Solutions

Hongying Liu, Hao Wang, Haoran Chu, Yibo Wu

摘要

An unsolved issue in widely used methods such as Support Vector Data Description (SVDD) and Small Sphere and Large Margin SVM (SSLM) for anomaly detection is their nonconvexity, which hampers the analysis of optimal solutions in a manner similar to SVMs and limits their applicability in large-scale scenarios. In this paper, we introduce a novel convex SSLM formulation which has been demonstrated to revert to a convex quadratic programming problem for hyperparameter values of interest. Leveraging the convexity of our method, we derive numerous results that are unattainable with traditional nonconvex approaches. We conduct a thorough analysis of how hyperparameters influence the optimal solution, pointing out scenarios where optimal solutions can be trivially found and identifying instances of illposedness. Most notably, we establish connections between our method and traditional approaches, providing a clear determination of when the optimal solution is unique-a task unachievable with traditional nonconvex methods. We also derive the $\nu$-property to elucidate the interactions between hyperparameters and the fractions of support vectors and margin errors in both positive and negative classes.

JMLR Diagnostics & Prediction · 发表于 2026

Three Types of Calibration using Properties and their Semantic and Formal Relationships

Rabanus Derr, Jessie Finocchiaro, Robert C. Williamson

摘要

Fueled by discussions around "trustworthiness" and algorithmic fairness, calibration of predictive systems has regained scholars' attention. The vanilla definition and understanding of calibration is, simply put, on all days on which the rain probability has been predicted to be $p$, the actual frequency of rain days was $p$. However, the increased attention has led to an immense variety of new notions of "calibration". Some of the notions are incomparable, serve different purposes, or imply each other. In this work, we provide two accounts which motivate calibration: self-realization of forecasted properties and precise estimation of incurred losses of the decision makers relying on forecasts. We substantiate the former via the reflection principle and the latter by actuarial fairness. For both accounts we formulate prototypical definitions via properties $\Gamma$ of outcome distributions, e.g., the mean or median. The prototypical definition for self-realization, which we call $\Gamma$-calibration, is equivalent to a certain type of swap regret under certain conditions. These implications are strongly connected to the omniprediction learning paradigm. The prototypical definition for precise loss estimation is a modification of decision calibration adopted from Zhao et al., 2021. For binary outcome sets both prototypical definitions coincide under appropriate choices of reference properties. For higher-dimensional outcome sets, both prototypical definitions can be subsumed by a natural extension of the binary definition, called distribution calibration with respect to a property. We conclude by commenting on the role of groupings in both accounts of calibration often used to obtain multicalibration. In sum, this work provides a semantic map of calibration in order to navigate a fragmented terrain of notions and definitions.

JMLR Bayesian Methods Machine Learning Networks & Graphs · 发表于 2026

The surrogate Gibbs-posterior of a corrected stochastic MALA: Towards uncertainty quantification for neural networks

Sebastian Bieringer, Gregor Kasieczka, Maximilian F. Steffen, Mathias Trabs

摘要

MALA is a popular gradient-based Markov chain Monte Carlo method to access the Gibbs-posterior distribution. Stochastic MALA (sMALA) scales to large data sets, but changes the target distribution from the Gibbs-posterior to a surrogate posterior which only exploits a reduced sample size. We introduce a corrected stochastic MALA (csMALA) with a simple correction term for which distance between the resulting surrogate posterior and the original Gibbs-posterior decreases in the full sample size while retaining scalability. In a nonparametric regression model, we prove a PAC-Bayes oracle inequality for the surrogate posterior. Uncertainties can be quantified by sampling from the surrogate posterior. Focusing on Bayesian neural networks, we analyze the diameter and coverage of credible balls for shallow neural networks and we show optimal contraction rates for deep neural networks. Our credibility result is independent of the correction and can also be applied to the standard Gibbs-posterior. A simulation study in a high-dimensional parameter space demonstrates that an estimator drawn from csMALA based on its surrogate Gibbs-posterior indeed exhibits these advantages in practice.

JMLR Other · 发表于 2026

The Within-Orbit Adaptive Leapfrog No-U-Turn Sampler

Nawaf Bou-Rabee, Bob Carpenter, Tore Selland Kleppe, Sifan Liu

摘要

Locally adapting parameters within Markov chain Monte Carlo methods while preserving reversibility is notoriously difficult. The success of the No-U-Turn Sampler (NUTS) largely stems from its clever local adaptation of the integration time in Hamiltonian Monte Carlo via a geometric U-turn condition. However, posterior distributions frequently exhibit multiscale geometries with extreme variations in scale, making it necessary to also adapt the leapfrog integrator's step size locally and dynamically. Despite its practical importance, this problem has remained largely open since the introduction of NUTS by Hoffman and Gelman (2014). To address this issue, we introduce the Within-Orbit Adaptive Leapfrog No-U-Turn Sampler (WALNUTS), a generalization of NUTS that adapts the leapfrog step size at fixed intervals of simulated time as the orbit evolves. At each interval, the algorithm selects the largest step size from a dyadic schedule that keeps the energy error below a user-specified threshold. Like NUTS, WALNUTS employs biased progressive state selection to favor states with positions that are further from the initial point along the orbit. Empirical evaluations on multiscale target distributions, including Neal's funnel and the Stock-Watson stochastic volatility time-series model, demonstrate that WALNUTS achieves substantial improvements in sampling efficiency and robustness compared to NUTS.

JMLR Optimization & Learning Theory · 发表于 2026

The Sample Complexity of Parameter-Free Stochastic Convex Optimization

Jared Lawrence, Ari Kalinsky, Hannah Bradfield, Yair Carmon, Oliver Hinder

摘要

We study the sample complexity of stochastic convex optimization when problem parameters such as the distance to optimality and the Lipschitz constant are unknown. We pursue two strategies. First, we develop a reliable model selection method that avoids overfitting to the validation set. This method allows us to generically tune the learning rate of stochastic optimization methods to match the optimal known-parameter sample complexity up to $\log\log$ factors. Second, we develop a regularization-based method that is specialized to the case that only the distance to optimality is unknown. More specifically, it uses norm-regularized empirical risk minimization to estimate the distance to optimality to within a constant factor, allowing known-parameter stochastic optimization methods to achieve optimal sample complexity. This method provides perfect adaptability to unknown distance to optimality, demonstrating a separation between the sample and computational complexity of parameter-free stochastic convex optimization. Combining these two methods allows us to simultaneously adapt to multiple problem structures. Experiments performing few-shot learning on CIFAR-10 by fine-tuning CLIP models and prompt engineering Gemini to count shapes indicate that our reliable model selection method can help mitigate overfitting to small validation sets.

Econometrica Econometrics & Economics · 发表于 2026,收录于 2026-10-20

The Role of Trade in Premature Deindustrialization and Industry Polarization

Michael Sposi, Kei-Mu Yi, Jing Zhang

摘要

We document two key patterns of structural change: premature deindustrialization and rising industry polarization. Using a dynamic open‐economy model with sector‐biased technological change (SBTC), we evaluate the role of trade in these shifts. Trade alone does not cause premature deindustrialization, but its interaction with SBTC—transmitting global technological change, amplifying the decline in manufacturing's relative price over time, and generating manufacturing trade imbalances across countries—deepens premature deindustrialization over time. Industry polarization, by contrast, emerges only through trade‐driven specialization, as countries diverge along lines of comparative advantage.

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 39 40 41 42 43 44 页