研究园简易版 · 经典书籍 · 期刊文章 · 有原文 · 完整版(可多选筛选、搜索) · 首页
2025 年的期刊文章 · 第 29 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 →
Econometrica Spatial & Time Series · 发表于 2025
Charles Hodgson, Gregory Lewis
摘要
We develop and estimate a model of consumer search with spatial learning. Consumers make inferences from previously searched objects to unsearched objects that are nearby in attribute space, generating path dependence in search sequences. The estimated model rationalizes patterns in data on online consumer search paths: search tends to converge to the chosen product in attribute space, and consumers take larger steps away from rarely purchased products. Eliminating spatial learning reduces consumer welfare by 12%: cross‐product inferences allow consumers to locate better products in a shorter time. Spatial learning has important implications for product recommendations on retail platforms. We show that consumer welfare can be reduced by unrepresentative product recommendations and that consumer‐optimal product recommendations depend on both consumer learning and competition between platforms.
Econometrica Other · 发表于 2025
Scott Kim, Petra Moser
摘要
This paper investigates how children affect women in science, using biographies in the American Men of Science (MoS 1956), linked with publications. First, we show that mothers have a unique life cycle pattern of productivity: While other scientists peak in their mid‐30s, mothers become less productive at that age and reach peak productivity in their early‐40s. Next, we estimate event studies of marriage, comparing mothers and fathers with other married scientists. Event study estimates show that the productivity of mothers declines until children reach school age, while fathers experience no change. These differences have important implications for tenure and participation: Just 27% of mothers achieve tenure, compared with 48% of fathers and 46% of other women. When women carried the full burden of childcare, the time costs of raising the baby boom led to a great loss of female scientists.
Econometrica Other · 发表于 2025
Juan Camilo Castillo
摘要
New technologies have recently led to a boom in real‐time pricing. I study the most salient example, surge pricing in ride hailing. Using data from Uber, I develop an empirical model of spatial equilibrium to measure the welfare effects of surge pricing. The model is composed of demand, supply, and a matching technology. It allows for temporal and spatial heterogeneity as well as randomness in supply and demand. I find that, relative to a uniform pricing counterfactual in which Uber sets the overall price level, surge pricing increases total welfare by 2.15% of gross revenue. Welfare effects differ substantially across sides of the market: rider surplus increases by 3.57% of gross revenue, whereas driver surplus and the platform's current profits decrease by 0.98% and 0.50% of gross revenue, respectively. Riders at all income levels benefit. Among drivers, those who work long hours are hurt the most, especially women.
JMLR Optimization & Learning Theory Genetics & Omics · 发表于 2025
Haoshu Xu, Hongzhe Li
摘要
This paper addresses regression analysis for covariance matrix-valued outcomes with Euclidean covariates, motivated by applications in single-cell genomics and neuroscience where covariance matrices are observed across many samples. Our analysis leverages Fréchet regression on the Bures-Wasserstein manifold to estimate the conditional Fréchet mean given covariates $x$. We establish a non-asymptotic uniform $\sqrt{n}$-rate of convergence (up to logarithmic factors) over covariates with $\|x\| \lesssim \sqrt{\log n}$ and derive a pointwise central limit theorem to enable statistical inference. For testing covariate effects, we devise a novel test whose null distribution converges to a weighted sum of independent chi-square distributions, with power guarantees against a sequence of contiguous alternatives. Simulations validate the accuracy of the asymptotic theory. Finally, we apply our methods to a single-cell gene expression dataset, revealing age-related changes in gene co-expression networks.
JMLR Machine Learning Optimization & Learning Theory · 发表于 2025
Xuefeng Gao, Hoang M. Nguyen, Lingjiong Zhu
摘要
Score-based generative models are a recent class of deep generative models with state-of-the-art performance in many applications. In this paper, we establish convergence guarantees for a general class of score-based generative models in the 2-Wasserstein distance, assuming accurate score estimates and smooth log-concave data distribution. We specialize our results to several concrete score-based generative models with specific choices of forward processes modeled by stochastic differential equations, and obtain an upper bound on the iteration complexity for each model, which demonstrates the impacts of different choices of the forward processes. We also provide a lower bound when the data distribution is Gaussian. Numerically, we experiment with score-based generative models with different forward processes for unconditional image generation on CIFAR-10. We find that the experimental results are in good agreement with our theoretical predictions on the iteration complexity.
JMLR Other · 发表于 2025
Pablo Badilla, Felipe Bravo-Marquez, María José Zambrano, Jorge Pérez
摘要
Word embeddings, which are a mapping of words into continuous vectors, are widely used in modern Natural Language Processing (NLP) systems. However, they are prone to inherit stereotypical social biases from the corpus on which they are built. The research community has focused on two main tasks to address this problem: 1) how to measure these biases, and 2) how to mitigate them. Word Embedding Fairness Evaluation (WEFE) is an open source library that implements many fairness metrics and mitigation methods in a unified framework. It also provides a standard interface for designing new ones. The software follows the object-oriented paradigm with a strong focus on extensibility. Each of its methods is appropriately documented, verified and tested. WEFE is not limited to just a library: it also contains several replications of previous studies as well as tutorials that serve as educational material for newcomers to the field. It is licensed under BSD-3 and can be easily installed through pip and conda package managers.
JMLR Econometrics & Economics Bayesian Methods · 发表于 2025
Charles C. Margossian, Loucas Pillaud-Vivien, Lawrence K. Saul
摘要
Given an intractable distribution $p$, the problem of variational inference (VI) is to find the best approximation from some more tractable family $Q$. Commonly, one chooses $Q$ to be a family of factorized distributions (i.e., the mean-field assumption), even though $p$ itself does not factorize. We show that this mismatch leads to an impossibility theorem: if $p$ does not factorize, then any factorized approximation $q\!\in\!Q$ can correctly estimate at most one of the following three measures of uncertainty: (i) the marginal variances, (ii) the marginal precisions, or (iii) the generalized variance (which for elliptical distributions is closely related to the entropy). In practice, the best variational approximation in $Q$ is found by minimizing some divergence $D(q,p)$ between distributions, and so we ask: how does the choice of divergence determine which measure of uncertainty, if any, is correctly estimated by VI? We consider the classic Kullback-Leibler divergences, the more general $\alpha$-divergences, and a score-based divergence which compares $\nabla \log p$ and $\nabla \log q$. We thoroughly analyze the case where $p$ is a Gaussian and $q$ is a (factorized) Gaussian. In this setting, we show that all the considered divergences can be ordered based on the estimates of uncertainty they yield as objective functions for VI. Finally, we empirically evaluate the validity of this ordering when the target distribution $p$ is not Gaussian.
JMLR Nonparametric & Semiparametric · 发表于 2025
Geoffrey Wolfer, Pierre Alquier
摘要
An important feature of kernel mean embeddings (KME) is that the rate of convergence of the empirical KME to the true distribution KME can be bounded independently of the dimension of the space, properties of the distribution and smoothness features of the kernel. We show how to speed-up convergence by leveraging variance information in the reproducing kernel Hilbert space. Furthermore, we show that even when such information is a priori unknown, we can efficiently estimate it from the data, recovering the desiderata of a distribution agnostic bound that enjoys acceleration in fortuitous settings. We further extend our results from independent data to stationary mixing sequences and illustrate our methods in the context of hypothesis testing and robust parametric estimation.
JMLR Other · 发表于 2025
Quoc Tran-Dinh
摘要
We develop two Variance-reduced Fast Operator Splitting Algorithms (VFOSA) to approximate solutions for a class of generalized equations, covering fundamental problems such as minimization, minimax problems, and variational inequalities as special cases. Our approach integrates recent advances in accelerated operator splitting and fixed-point methods, co-hypomonotonicity structure, and variance reduction techniques. First, we introduce a class of variance-reduced estimators and establish their variance-reduction bounds. This class includes both unbiased and biased instances and comprises common estimators as special cases, including SVRG, SAGA, SARAH, and Hybrid-SGD. Second, we design a novel accelerated variance-reduced forward-backward splitting (FBS) method using these estimators to solve generalized equations in both finite-sum and expectation settings. Our algorithm achieves both O(1/k^2) and o(1/k^2) convergence rates on the expected squared norm E[ ||G_{\lambda}x^k||^2] of the FBS residual G_{\lambda}, where k is the iteration counter. Additionally, we establish almost sure convergence rates and the almost sure convergence of iterates to a solution of the underlying generalized equation. Unlike existing stochastic operator splitting algorithms, our methods accommodate co-hypomonotone operators, which can include nonmonotone problems arising in recent applications. Third, we specify our method for each concrete estimator mentioned above and derive the corresponding oracle complexity, demonstrating that these variants achieve the best-known oracle complexity bounds without requiring additional enhancement techniques. Fourth, we develop a variance-reduced fast backward-forward splitting (BFS) method, which attains similar convergence results and oracle complexity bounds as our FBS-based algorithm. Finally, we validate our results through numerical experiments and compare their performance with existing methods.
SMMR Clinical Trials Diagnostics & Prediction · 发表于 2025
Xijin Chen, Pavel Mozgunov, Richard D Baird, Thomas Jaki
JMLR Causal Inference Networks & Graphs Diagnostics & Prediction · 发表于 2025
Brecht Verbeken, Marie-Anne Guerry, Wouter Verbeke, Sam Verboven
摘要
Uplift modelling is a subfield of causal learning that focuses on ranking entities by individual treatment effects. Uplift models are typically evaluated using Qini curves or Qini scores. While intuitive, the theoretical grounding for Qini in the literature is limited, and the mathematical connection to the well-understood Receiver Operating Characteristic (ROC) curve is unclear. In this paper, we introduce pROCini, a novel uplift evaluation metric that improves upon Qini in two important ways. First, it explicitly incorporates more information by taking into account negative outcomes. Second, it leverages this additional information within the Ordinal Dominance Graph framework, which is the basis behind the well known ROC curve, resulting in a mathematically well-behaved metric that facilitates theoretical analysis. We derive confidence bounds for pROCini, exploiting its theoretical properties. Finally, we empirically validate the improved discriminative power of ROCini and pROCini in a simulation study as well as via experiments on real data.
JMLR Nonparametric & Semiparametric · 发表于 2025
Parthe Pandit, Zhichao Wang, Yizhe Zhu
摘要
Kernel ridge regression (KRR) is a popular class of machine learning models that has become an important tool for understanding deep learning. Much of the focus thus far has been on studying the proportional asymptotic regime, $n \asymp d$, where $n$ is the number of training samples and $d$ is the dimension of the dataset. In the proportional regime, under certain conditions on the data distribution, the kernel random matrix involved in KRR exhibits behavior akin to that of a linear kernel. In this work, we extend the study of kernel regression to the quadratic asymptotic regime, where $n \asymp d^2$. In this regime, we demonstrate that a broad class of inner-product kernels exhibits behavior similar to a quadratic kernel. Specifically, we establish an operator norm approximation bound for the difference between the original kernel random matrix and a quadratic kernel random matrix with additional correction terms compared to the Taylor expansion of the kernel functions. The approximation works for general data distributions under a Gaussian-moment-matching assumption with a covariance structure. This new approximation is utilized to obtain a limiting spectral distribution of the original kernel matrix and characterize the precise asymptotic training and test errors for KRR in the quadratic regime when $n/d^2$ converges to a non-zero constant. The generalization errors are obtained for (i) a random teacher model, (ii) a deterministic teacher model where the weights are perfectly aligned with the covariance of the data. Under the random teacher model setting, we also verify that the generalized cross-validation (GCV) estimator can consistently estimate the generalization error in the quadratic regime for anisotropic data. Our proof techniques combine moment methods, Wick's formula, orthogonal polynomials, and resolvent analysis of random matrices with correlated entries.
JMLR Optimization & Learning Theory · 发表于 2025
Lijun Zhang, Yibo Wang, Guanghui Wang, Jinfeng Yi, Tianbao Yang
摘要
Recently, several universal methods have been proposed for online convex optimization, and attain minimax rates for multiple types of convex functions simultaneously. However, they need to design and optimize one surrogate loss for each type of functions, making it difficult to exploit the structure of the problem and utilize existing algorithms. In this paper, we propose a simple strategy for universal online convex optimization, which avoids these limitations. The key idea is to construct a set of experts to process the original online functions, and deploy a meta-algorithm over the linearized losses to aggregate predictions from experts. Specifically, the meta-algorithm is required to yield a second-order bound with excess losses, so that it can leverage strong convexity and exponential concavity to control the meta-regret. In this way, our strategy inherits the theoretical guarantee of any expert designed for strongly convex functions and exponentially concave functions, up to a double logarithmic factor. As a result, we can plug in off-the-shelf online solvers as black-box experts to deliver problem-dependent regret bounds. For general convex functions, it maintains the minimax optimality and also achieves a small-loss bound. Furthermore, we extend our universal strategy to online composite optimization, where the loss function comprises a time-varying function and a fixed regularizer. To deal with the composite loss functions, we employ a meta-algorithm based on the optimistic online learning framework, which not only enjoys a second-order bound, but also can utilize estimations for upcoming loss functions. With suitable configurations, we show that the additional regularizer does not contribute to the meta-regret, thus ensuring the universality in the composite setting.
Econometrica Bayesian Methods · 发表于 2025
Jonas E. Arias, Juan F. Rubio-Ramírez, Daniel F. Waggoner
摘要
There has been a call for caution regarding the standard procedure for Bayesian inference in set‐identified structural vector autoregressions on the grounds that the common practice of using a uniform prior over the set of orthogonal matrices induces a non‐uniform prior for individual impulse responses or other quantities of interest. This paper challenges this call by formally showing that when the focus is on joint inference, the uniform prior over the set of orthogonal matrices is not only sufficient but also necessary for inference based on a uniform joint prior distribution over the identified set for the vector of impulse responses. In addition, we show how to conduct inference based on a uniform joint prior distribution for the vector of impulse responses.
JMLR Other · 发表于 2025
Lingxiao Zhao, Xueying Ding, Lijun Yu, Leman Akoglu
摘要
Discrete diffusion models have attracted significant attention for their application to naturally discrete data, such as language and graphs. While discrete-time discrete diffusion has been established for some time, it was only recently that Campbell et al. (2022) introduced the first framework for continuous-time discrete diffusion. However, their training and backward sampling processes significantly differ from those of the discrete-time version, requiring nontrivial approximations for tractability. In this paper, we first introduce a series of generalizations and simplifications of the evidence lower bound (ELBO) that facilitate more accurate and easier optimization both discrete- and continuous-time discrete diffusion. We further establish a unification of discrete- and continuous-time discrete diffusion through shared forward process and backward parameterization. Thanks to this unification, the continuous-time diffusion can now utilize the exact and efficient backward process developed for the discrete-time case, avoiding the need for costly and inexact approximations. Similarly, the discrete-time diffusion now also employ the MCMC corrector, which was previously exclusive to the continuous-time case. Extensive experiments and ablations demonstrate the significant improvement, and we open-source our code at: https://github.com/LingxiaoShawn/USD3.
JMLR Networks & Graphs · 发表于 2025
Peng Wang, Xiao Li, Can Yaras, Zhihui Zhu, Laura Balzano, Wei Hu, Qing Qu
摘要
Over the past decade, deep learning has proven to be a highly effective tool for learning meaningful features from raw data. However, it remains an open question how deep networks perform hierarchical feature learning across layers. In this work, we attempt to unveil this mystery by investigating the structures of intermediate features. Motivated by our empirical findings that linear layers mimic the roles of deep layers in nonlinear networks for feature learning, we explore how deep linear networks transform input data into output by investigating the output (i.e., features) of each layer after training in the context of multi-class classification problems. Toward this goal, we first define metrics to measure within-class compression and between-class discrimination of intermediate features, respectively. Through theoretical analysis of these two metrics, we show that the evolution of features follows a simple and quantitative pattern from shallow to deep layers when the input data is nearly orthogonal and the network weights are minimum-norm, balanced, and approximately low-rank: each layer of the linear network progressively compresses within-class features at a geometric rate and discriminates between-class features at a linear rate with respect to the number of layers that data have passed through. To the best of our knowledge, this is the first quantitative characterization of feature evolution in hierarchical representations of deep linear networks. Moreover, our extensive experiments not only validate our theoretical results but also reveal a similar pattern in deep nonlinear networks, which aligns well with recent empirical studies. Finally, we demonstrate the practical value of our results in transfer learning.
JMLR Other · 发表于 2025
Shayan Hundrieser, Florian Heinemann, Marcel Klatt, Marina Struleva, Axel Munk
摘要
We analyze statistical properties of plug-in estimators for unbalanced optimal transport quantities between finitely supported measures in different prototypical sampling models. Specifically, our main results provide non-asymptotic bounds on the expected error of empirical Kantorovich-Rubinstein (KR) distance, plans, and barycenters for mass penalty parameter $C>0$. The impact of the mass penalty parameter $C$ is studied in detail. Based on this analysis, we mathematically justify randomized computational schemes for KR quantities which can be used for fast approximate computations in combination with any exact solver. Using synthetic and real datasets, we empirically analyze the behavior of the expected errors in simulation studies and illustrate the validity of our theoretical bounds.
JMLR Optimization & Learning Theory · 发表于 2025
Tianyi Lin, Chi Jin, Michael I. Jordan
摘要
We provide a unified analysis of two-timescale gradient descent ascent (TTGDA) for solving structured nonconvex minimax optimization problems in the form of $\min_x \max_{y \in Y} f(x, y)$, where the objective function $f(x, y)$ is nonconvex in $x$ and concave in $y$, and the constraint set $Y \subseteq \mathbb{R}^n$ is convex and bounded. In the convex-concave setting, the single-timescale gradient descent ascent (GDA) algorithm is widely used in applications and has been shown to have strong convergence guarantees. In more general settings, however, it can fail to converge. Our contribution is to design TTGDA algorithms that are effective beyond the convex-concave setting, efficiently finding a stationary point of the function $\Phi(\cdot) := \max_{y \in Y} f(\cdot, y)$. We also establish theoretical bounds on the complexity of solving both smooth and nonsmooth nonconvex-concave minimax optimization problems. To the best of our knowledge, this is the first systematic analysis of TTGDA for nonconvex minimax optimization, shedding light on its superior performance in training generative adversarial networks (GANs) and in other real-world application problems.
Econometrica Other · 发表于 2025
Adam Kapor
摘要
Transparency versus opacity is an important dimension of college admission policy. Colleges may gain useful information from a holistic review of applicants' materials, but in doing so may contribute to uncertainty that discourages potential applicants with poor information. This paper investigates the impacts of admissions transparency in the context of Texas' Top Ten Percent Plan, using survey and administrative data from Texas and a model of college applications, admissions, enrollment, grades, and persistence. I estimate that two thirds of the plan's 9.1 point impact on top‐decile students' probability of attending a flagship university was due to information rather than mechanical effects. Students induced to enroll are more likely to come from low‐income high schools, and academically outperform the students that they displace. These effects would be larger if complemented by financial‐aid information, and are driven by transparency, not misalignment between the rules used for automatic and discretionary admissions.
JMLR Machine Learning Networks & Graphs · 发表于 2025
Qitian Wu, David Wipf, Junchi Yan
摘要
Learning representations for structured data with certain geometries (e.g., observed or unobserved) is a fundamental challenge, wherein message passing neural networks (MPNNs) have become a de facto class of model solutions. In this paper, inspired by physical systems, we propose an energy-constrained diffusion model, which integrates the inductive bias of diffusion on manifolds with layer-wise constraints of energy minimization. We identify that the diffusion operators have a one-to-one correspondence with the energy functions implicitly descended by the diffusion process, and the finite-difference iteration for solving the energy-constrained diffusion system induces the propagation layers of various types of MPNNs operating on observed or latent structures. This leads to a unified mathematical framework for common neural architectures whose computational flows can be cast as message passing (or its special case), including MLPs, GNNs, and Transformers. Building on these insights, we devise a new class of neural message passing models, dubbed diffusion-inspired Transformers (DIFFormer), whose global attention layers are derived from the principled energy-constrained diffusion framework. Across diverse datasets ranging from real-world networks to images, texts, and physical particles, we demonstrate that the new model achieves promising performance in scenarios where the data structures are observed (as a graph), partially observed, or entirely unobserved.
JMLR Nonparametric & Semiparametric · 发表于 2025
Xavier Emery, Emilio Porcu, Moreno Bevilacqua
摘要
There exists a plethora of parametric models for positive definite kernels in Euclidean spaces, and their use is ubiquitous in statistics, machine learning, numerical analysis, and approximation theory. Usually, the kernel parameters index certain features of an associated process. Amongst those features, smoothness (in the sense of Sobolev spaces, mean square differentiability, and fractal dimensions), compact or global supports, and negative dependencies (hole effects) are of interest to several theoretical and applied disciplines. This paper unifies a wealth of well-known kernels into a single parametric class that encompasses them as special cases, attained either by exact parameterization or through parametric asymptotics. We furthermore find parametric restrictions under which we can characterize the Sobolev space that is norm equivalent to the RKHS associated with the new kernel. As a by-product, we infer the Sobolev spaces that are associated with existing classes of kernels. We illustrate the main properties of the new class, show how this class can switch from compact to global supports, and provide special cases for which the kernel attains negative values over nontrivial intervals. Hence, the proposed class of kernel is the reproducing kernel of a Hilbert space that contains many special cases, including the celebrated Matérn and Wendland kernels, as well as their aliases with hole effects.
JMLR Machine Learning Networks & Graphs · 发表于 2025
Akshay Kumar, Jarvis Haupt
摘要
Recent works exploring the training dynamics of homogeneous neural network weights under gradient flow with small initialization have established that in the early stages of training, the weights remain small and near the origin, but converge in direction. Building on this, the current paper studies the gradient flow dynamics of homogeneous neural networks with locally Lipschitz gradients, after they escape the origin. Insights gained from this analysis are used to characterize the first saddle point encountered by gradient flow after escaping the origin. Also, it is shown that for homogeneous feed-forward neural networks, under certain conditions, the sparsity structure emerging among the weights before the escape is preserved after escaping the origin and until reaching the next saddle point.
JMLR Networks & Graphs Machine Learning · 发表于 2025
Brendon G. Anderson, Ziye Ma, Jingqi Li, Somayeh Sojoudi
摘要
In this paper, we study certifying the robustness of ReLU neural networks against adversarial input perturbations. To diminish the relaxation error suffered by the popular linear programming (LP) and semidefinite programming (SDP) certification methods, we take a branch-and-bound approach to propose partitioning the input uncertainty set and solving the relaxations on each part separately. We show that this approach reduces relaxation error, and that the error is eliminated entirely upon performing an LP relaxation with a partition intelligently designed to exploit the nature of the ReLU activations. To scale this approach to large networks, we consider using a coarser partition whereby the number of parts in the partition is reduced. We prove that computing such a coarse partition that directly minimizes the LP relaxation error is NP-hard. By instead minimizing the worst-case LP relaxation error, we develop a closed-form branching scheme in the single-hidden layer case. We extend the analysis to the SDP, where the feasible set geometry is exploited to design a branching scheme that minimizes the worst-case SDP relaxation error. Experiments on MNIST, CIFAR-10, and Wisconsin breast cancer diagnosis classifiers demonstrate significant increases in the percentages of test samples certified. By independently increasing the input size and the number of layers, we empirically illustrate under which regimes the branched LP and branched SDP are best applied. Finally, we extend our LP branching method into a multi-layer branching heuristic, which attains comparable performance to prior state-of-the-art heuristics on large-scale, deep neural network certification benchmarks.
JMLR Testing & Uncertainty Machine Learning Networks & Graphs · 发表于 2025
Jianguo Huang, Jianqing Song, Xuanning Zhou, Bingyi Jing, Hongxin Wei
摘要
Conformal prediction (CP) is a powerful statistical framework that generates prediction intervals or sets with guaranteed coverage probability. While CP algorithms have evolved beyond traditional classifiers and regressors to sophisticated deep learning models like deep neural networks (DNNs), graph neural networks (GNNs), and large language models (LLMs), existing CP libraries often lack the model support and scalability for large-scale deep learning (DL) scenarios. This paper introduces TorchCP, a PyTorch-native library designed to integrate state-of-the-art CP algorithms into DL techniques, including DNN-based classifiers/regressors, GNNs, and LLMs. Released under the LGPL-3.0 license, TorchCP comprises about 16k lines of code, validated with 100% unit test coverage and detailed documentation. Notably, TorchCP enables CP-specific training algorithms, online prediction, and GPU-accelerated batch processing, achieving up to 90% reduction in inference time on large datasets. With its low-coupling design, comprehensive suite of advanced methods, and full GPU scalability, TorchCP empowers researchers and practitioners to enhance uncertainty quantification across cutting-edge applications.
JMLR Optimization & Learning Theory Networks & Graphs · 发表于 2025
Martin Bauer, Facundo Mémoli, Tom Needham, Mao Nishino
摘要
The Gromov-Wasserstein (GW) distance is a powerful tool for comparing metric measure spaces which has found broad applications in data science and machine learning. Driven by the need to analyze data sets whose objects have increasingly complex structure (such as node and edge-attributed graphs), several variants of GW distance have been introduced in the recent literature. With a view toward establishing a general framework for the theory of GW-like distances, this paper considers a vast generalization of the notion of a metric measure space: for an arbitrary metric space $Z$, we define a $Z$-network to be a measure space endowed with a kernel valued in $Z$. We introduce a method for comparing $Z$-networks by defining a generalization of GW distance, which we refer to as $Z$-Gromov-Wasserstein ($Z$-GW) distance. This construction subsumes many previously known metrics and offers a unified approach to understanding their shared properties. This paper demonstrates that the $Z$-GW distance defines a metric on the space of $Z$-networks which retains desirable properties of $Z$, such as separability, completeness, and geodesicity. Many of these properties were unknown for existing variants of GW distance that fall under our framework. Our focus is on foundational theory, but our results also include computable lower bounds and approximations of the distance which will be useful for practical applications.
Econometrica Econometrics & Economics · 发表于 2025
Eliana Carranza, Aletheia Donald, Florian Grosset-Touba, Supreet Kaur
摘要
In low‐income communities in both rich and poor countries, redistributive transfers within kin and social networks are frequent. Such arrangements may distort labor supply—acting as a “social tax” that dampens the incentive to work. We document that across countries, from the United States to Côte d'Ivoire, low‐income groups report strong pressure to share earned income with others; in addition, social groups that undertake more interpersonal transfers work fewer hours. Using a field experiment, we enable piece‐rate factory workers in Côte d'Ivoire to shield income using blocked savings accounts over 9 months. Workers may only deposit earnings increases, relative to baseline, mitigating income effects on labor supply. Offering Private accounts raises work attendance by 6.5% and earnings by 9.4%. These treatment effects are concentrated among workers who report higher redistributive pressure at baseline. To obtain further suggestive evidence on mechanisms, in a supplementary experiment, we vary whether blocked accounts are private or known to the worker's network. When accounts are private, take‐up is substantively higher (60% vs. 14%), with a resultant 8.8% higher earnings. Outgoing transfers do not decline, indicating no loss in redistribution. The welfare benefits of informal redistribution may come at a cost, depressing labor supply and productivity.
Econometrica Econometrics & Economics · 发表于 2025
S. Nageeb Ali, Maximilian Mihm, Lucas Siga
摘要
This paper offers a strategic rationale for zero‐sum thinking in elections. We show that asymmetric information and distributional considerations together make voters wary of policies supported by others. This force impels a majority of voters to support policies contrary to their preferences and information. Our analysis identifies and interprets a form of “adverse correlation” that is necessary and sufficient for zero‐sum thinking to prevail in equilibrium.
JMLR Machine Learning · 发表于 2025
Shuze Daniel Liu, Shuhang Chen, Shangtong Zhang
摘要
Stochastic approximation is a class of algorithms that update a vector iteratively, incrementally, and stochastically, including, e.g., stochastic gradient descent and temporal difference learning. One fundamental challenge in analyzing a stochastic approximation algorithm is to establish its stability, i.e., to show that the stochastic vector iterates are bounded almost surely. In this paper, we extend the celebrated Borkar-Meyn theorem for stability from the Martingale difference noise setting to the Markovian noise setting, which greatly improves its applicability in reinforcement learning, especially in those off-policy reinforcement learning algorithms with linear function approximation and eligibility traces. Central to our analysis is the diminishing asymptotic rate of change of a few functions, which is implied by both a form of the strong law of large numbers and a form of the law of the iterated logarithm.
Econometrica Econometrics & Economics · 发表于 2025
Ana Cecília Fieler, Jonathan Eaton
摘要
Welfare depends on the quantity, quality, and range of goods consumed. We use trade data, which report the quantities and prices of the individual goods that countries exchange, to learn about how the gains from trade and growth break down into these different margins. Our general equilibrium model, in which both quality and quantity contribute to consumption and to production, captures (i) how prices increase with importer and exporter per capita income, (ii) how the range of goods traded rises with importer and exporter size, and (iii) how products traveling longer distances have higher prices. Our framework can deliver a standard gravity formulation for total trade flows and for the gains from trade. We find that growth in the extensive margin contributes to about half of overall gains. Quality plays a larger role in the welfare gains from international trade than from economic growth due to selection.
Econometrica Econometrics & Economics · 发表于 2025
Andrew Garin, Dmitri Koustas, Carl McPherson, Samuel Norris, Matthew Pecenco, Evan K. Rose, Yotam Shem-Tov, Jeffrey Weaver
摘要
We study the effect of incarceration on wages, self‐employment, and taxes and transfers in North Carolina and Ohio using two quasi‐experimental research designs: discontinuities in sentencing guidelines and random assignment to judges. Across both states, incarceration generates short‐term drops in economic activity while individuals remain in prison. As a result, a year‐long sentence decreases cumulative earnings over five years by 13%. Beyond five years, however, there is no evidence of lower employment, wage earnings, or self‐employment in either state, as well as among defendants with no prior incarceration history. These results suggest that upstream factors, such as other types of criminal justice interactions or pre‐existing labor market detachment, are more likely to be the cause of low earnings among the previously incarcerated, who we estimate would earn just $5000 per year on average if spared a prison sentence.
JMLR Nonparametric & Semiparametric Networks & Graphs · 发表于 2025
Peter Mostowsky, Vincent Dutordoir, Iskander Azangulov, Noémie Jaquier, Michael John Hutchinson, Aditya Ravuri, Leonel Rozo, Alexander Terenin, Viacheslav Borovitskiy
摘要
Kernels are a fundamental technical primitive in machine learning. In recent years, kernel-based methods such as Gaussian processes are becoming increasingly important in applications where quantifying uncertainty is of key interest. In settings that involve structured data defined on graphs, meshes, manifolds, or other related spaces, defining kernels with good uncertainty-quantification behavior, and computing their value numerically, is less straightforward than in the Euclidean setting. To address this difficulty, we present GeometricKernels, a Python software package which implements the geometric analogs of classical Euclidean squared exponential--also known as heat--and Matérn kernels, which are widely-used in settings where uncertainty is of key interest. As a byproduct, we obtain the ability to compute Fourier-feature-type expansions, which are widely used in their own right, on a wide set of geometric spaces. Our implementation supports automatic differentiation in every major current framework simultaneously via a backend-agnostic design. In this companion paper to the package and its documentation, we outline the capabilities of the package and present an illustrated example of its interface. We also include a brief overview of the theory the package is built upon and provide some historic context in the appendix.
JMLR High-Dimensional Optimization & Learning Theory · 发表于 2025
Nikhil Ghosh, Spencer Frei, Wooseok Ha, Bin Yu
摘要
In this work, we investigate the dynamics of stochastic gradient descent (SGD) when training a single-neuron autoencoder with linear or ReLU activation on orthogonal data. We show that for this non-convex problem, randomly initialized SGD with a constant step size successfully finds a global minimum for any batch size choice. However, the particular global minimum found depends upon the batch size. In the full-batch setting, we show that the solution is dense (i.e., not sparse) and is highly aligned with its initialized direction, showing that relatively little feature learning occurs. On the other hand, for any batch size strictly smaller than the number of samples, SGD finds a global minimum that is sparse and nearly orthogonal to its initialization, showing that the randomness of stochastic gradients induces a qualitatively different type of "feature selection" in this setting. Moreover, if we measure the sharpness of the minimum by the trace of the Hessian, the minima found with full-batch gradient descent are flatter than those found with strictly smaller batch sizes, in contrast to previous works which suggest that large batches lead to sharper minima. To prove convergence of SGD with a constant step size, we introduce a powerful tool from the theory of non-homogeneous random walks which may be of independent interest.
Econometrica Econometrics & Economics · 发表于 2025
Benjamin L. Collier, Cameron M. Ellis, Benjamin J. Keys
摘要
How do collateral requirements impact consumer borrowing behavior? Using administrative loan application and performance data from the U.S. Federal Disaster Loan Program, we exploit a loan amount threshold above which households must post their residence as collateral. Our bunching estimates suggest that the median borrower is willing to give up 40% of their loan amount to avoid posting collateral. Exploiting time variation in the threshold, we estimate collateral causally reduces default rates by 36%. Finally, we structurally estimate households' attachment to their homes, net of any equity, and find a median value of $11,000. Attachment creates a wedge between lender and borrower valuation of collateral of 15%. Our results explain high perceived default costs in the mortgage market, and document the importance of collateral for reducing moral hazard in consumer credit markets.
JMLR Meta-Analysis & Data Integration Optimization & Learning Theory · 发表于 2025
Jiin Woo, Gauri Joshi, Yuejie Chi
摘要
In this paper, we consider federated Q-learning, which aims to learn an optimal Q-function by periodically aggregating local Q-estimates trained on local data alone. Focusing on infinite-horizon tabular Markov decision processes, we provide sample complexity guarantees for both the synchronous and asynchronous variants of federated Q-learning, which exhibit a linear speedup with respect to the number of agents and near-optimal dependencies on other salient problem parameters. In the asynchronous setting, existing analyses of federated Q-learning, which adopt an equally weighted averaging of local Q-estimates, require that every agent covers the entire state-action space. In contrast, our improved sample complexity scales inverse proportionally to the minimum entry of the average stationary state-action occupancy distribution of all agents, thus only requiring the agents to collectively cover the entire state-action space, unveiling the blessing of heterogeneity. However, its sample complexity still suffers when the local trajectories are highly heterogeneous. In response, we propose a novel federated Q-learning algorithm with importance averaging, giving larger weights to more frequently visited state-action pairs, which achieves a robust linear speedup as if all trajectories are centrally processed, regardless of the heterogeneity of local behavior policies.
JMLR Other · 发表于 2025
Renhao Wang, Yu Sun, Arnuv Tandon, Yossi Gandelsman, Xinlei Chen, Alexei A. Efros, Xiaolong Wang
摘要
Prior work has established Test-Time Training (TTT) as a general framework to further improve a trained model at test time. Before making a prediction on each test instance, the model is first trained on the same instance using a self-supervised task such as reconstruction. We extend TTT to the streaming setting, where multiple test instances - video frames in our case - arrive in temporal order. Our extension is online TTT: The current model is initialized from the previous model, then trained on the current frame and a small window of frames immediately before. Online TTT significantly outperforms the fixed-model baseline for four tasks, on three real-world datasets. The improvements are more than 2.2x and 1.5x for instance and panoptic segmentation. Surprisingly, online TTT also outperforms its offline variant that accesses strictly more information, training on all frames from the entire test video regardless of temporal order. This finding challenges those in prior work using synthetic videos. We formalize a notion of locality as the advantage of online over offline TTT, and analyze its role with ablations and a theory based on bias-variance trade-off.
Econometrica Econometrics & Economics · 发表于 2025
Michael Weber, Bernardo Candia, Hassan Afrouzi, Tiziano Ropele, Rodrigo Lluberas, Serafin Frache, Brent Meyer, Saten Kumar, Yuriy Gorodnichenko, Dimitris Georgarakos, Olivier Coibion, Geoff Kenny, Jorge Ponce
摘要
Using randomized control trials (RCTs) applied over time in different countries, we study whether the economic environment affects how agents learn from new information. We show that as inflation rose in advanced economies, both households and firms became more attentive and informed about publicly available news about inflation, leading them to respond less to exogenously provided information about inflation and monetary policy. We also study the effects of RCTs in countries where inflation has been consistently high (Uruguay) and low (New Zealand) as well as what happens when the same agents are repeatedly provided information in both low‐ and high‐inflation environments (Italy). Our results broadly support models in which inattention is an endogenous outcome that depends on the economic environment.
JMLR Machine Learning · 发表于 2025
Si-Yang Liu, Hao-Run Cai, Qi-Le Zhou, Huai-Hong Yin, Tao Zhou, Jun-Peng Jiang, Han-Jia Ye
摘要
Tabular data is a prevalent source in machine learning. While classical methods have proven effective, deep learning methods for tabular data are emerging as flexible alternatives due to their capacity to uncover hidden patterns and capture complex interactions. Considering that deep tabular methods exhibit diverse design philosophies, including the ways they handle features, design learning objectives, and construct model architectures, we introduce Talent (Tabular Analytics and Learning Toolbox), a versatile toolbox for utilizing, analyzing, and comparing these methods. Talent includes over 35 deep tabular prediction methods, offering various encoding and normalization modules, all within a unified, easily extensible interface. We demonstrate its design, application, and performance evaluation in case studies. The code is available at https://github.com/LAMDA-Tabular/TALENT.
JMLR Other · 发表于 2025
Matteo Bettini, Ajay Shankar, Amanda Prorok
摘要
Evolutionary science provides evidence that diversity confers resilience in natural systems. Yet, traditional multi-agent reinforcement learning techniques commonly enforce homogeneity to increase training sample efficiency. When a system of learning agents is not constrained to homogeneous policies, individuals may develop diverse behaviors, resulting in emergent complementarity that benefits the system. Despite this, there is a surprising lack of tools that quantify behavioral diversity. Such techniques would pave the way towards understanding the impact of diversity in collective artificial intelligence and enabling its control. In this paper, we introduce System Neural Diversity (SND): a measure of behavioral heterogeneity in multi-agent systems. We discuss and prove its theoretical properties, and compare it with alternate, state-of-the-art behavioral diversity metrics used in the robotics domain. Through simulations of a variety of cooperative multi-robot tasks, we show how our metric constitutes an important tool that enables measurement and control of behavioral heterogeneity. In dynamic tasks, where the problem is affected by repeated disturbances during training, we show that SND allows us to measure latent resilience skills acquired by the agents, while other proxies, such as task performance (reward), fail to. Finally, we show how the metric can be employed to control diversity, allowing us to enforce a desired heterogeneity set-point or range. We demonstrate how this paradigm can be used to bootstrap the exploration phase, finding optimal policies faster, thus enabling novel and more efficient MARL paradigms.
JMLR Other · 发表于 2025
Verónica Álvarez, Santiago Mazuelas, Jose A. Lozano
摘要
Multiple supervised learning scenarios are composed by a sequence of classification tasks. For instance, multi-task learning and continual learning aim to learn a sequence of tasks that is either fixed or grows over time. Existing techniques for learning tasks that are in a sequence are tailored to specific scenarios, lacking adaptability to others. In addition, most of existing techniques consider situations in which the order of the tasks in the sequence is not relevant. However, it is common that tasks in a sequence are evolving in the sense that consecutive tasks often have a higher similarity. This paper presents a learning methodology that is applicable to multiple supervised learning scenarios and adapts to evolving tasks. Differently from existing techniques, we provide computable tight performance guarantees and analytically characterize the increase in the effective sample size. Experiments on benchmark datasets show the performance improvement of the proposed methodology in multiple scenarios and the reliability of the presented performance guarantees.
JMLR Latent Variables & Mixtures Causal Inference · 发表于 2025
Jeffrey Adams, Niels Richard Hansen
摘要
The deconfounder was proposed as a method for estimating causal parameters in a context with multiple causes and unobserved confounding. It is based on recovery of a latent variable from the observed causes. We disentangle the causal interpretation from the statistical estimation problem and show that the deconfounder in general estimates adjusted regression target parameters. It does so by outcome regression adjusted for the recovered latent variable termed the substitute. We refer to the general algorithm, stripped of causal assumptions, as substitute adjustment. We give theoretical results to support that substitute adjustment estimates adjusted regression parameters when the regressors are conditionally independent given the latent variable. We also introduce a variant of our substitute adjustment algorithm that estimates an assumption-lean target parameter with minimal model assumptions. We then give finite sample bounds and asymptotic results supporting substitute adjustment estimation in the case where the latent variable takes values in a finite set. A simulation study illustrates finite sample properties of substitute adjustment and shows that it can be a viable method for adjusted regression when the recovery error is small. Most importantly, we present clear assumptions about the data generating distribution that allow us to control the recovery error and ultimately the estimation error that can be attributed to the use of substitutes.