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
2022 年的期刊文章 · 第 36 页
共 1481 篇,按发表时间排,新的在前,每页 40 篇。这一页是 2022 到 2022。
按期刊、按主题看 →
JMLR Machine Learning · 发表于 2022
Shengyi Huang, Rousslan Fernand Julien Dossa, Chang Ye, Jeff Braga, Dipam Chakraborty, Kinal Mehta, João G.M. Araújo
摘要
CleanRL is an open-source library that provides high-quality single-file implementations of Deep Reinforcement Learning (DRL) algorithms. These single-file implementations are self-contained algorithm variant files such as dqn.py, ppo.py, and ppo_atari.py that individually include all algorithm variant's implementation details. Such a paradigm significantly reduces the complexity and the lines of code (LOC) in each implemented variant, which makes them quicker and easier to understand. This paradigm gives the researchers the most fine-grained control over all aspects of the algorithm in a single file, allowing them to prototype novel features quickly. Despite having succinct implementations, CleanRL's codebase is thoroughly documented and benchmarked to ensure performance is on par with reputable sources. As a result, CleanRL produces a repository tailor-fit for two purposes: 1) understanding all implementation details of DRL algorithms and 2) quickly prototyping novel features. CleanRL's source code can be found at https://github.com/vwxyzjn/cleanrl.
JMLR Spatial & Time Series Networks & Graphs Nonparametric & Semiparametric · 发表于 2022
Oscar Hernan Madrid Padilla, Yi Yu, Carey E. Priebe
摘要
In this paper, we study the offline change point localization problem in a sequence of dependent nonparametric random dot product graphs. To be specific, assume that at every time point, a network is generated from a nonparametric random dot product graph model (see e.g. Athreya et al., 2018), where the latent positions are generated from unknown underlying distributions. The underlying distributions are piecewise constant in time and change at unknown locations, called change points. Most importantly, we allow for dependence among networks generated between two consecutive change points. This setting incorporates edge-dependence within networks and temporal dependence between networks, which is the most flexible setting in the published literature. To accomplish the task of consistently localizing change points, we propose a novel change point detection algorithm, consisting of two steps. First, we estimate the latent positions of the random dot product model, our theoretical result being a refined version of the state-of-the-art results, allowing the dimension of the latent positions to diverge. Subsequently, we construct a nonparametric version of the CUSUM statistic (e.g. Page, 1954; Padilla et al., 2019a) that allows for temporal dependence. Consistent localization is proved theoretically and supported by extensive numerical experiments, which illustrate state-of-the-art performance. We also provide in depth discussion of possible extensions to give more understanding and insights.
Econometrica Causal Inference Econometrics & Economics · 发表于 2022
Guido W. Imbens
摘要
This essay describes the evolution and recent convergence of two methodological approaches to causal inference. The first one, in statistics, started with the analysis and design of randomized experiments. The second, in econometrics, focused on settings with economic agents making optimal choices. I argue that the local average treatment effects framework facilitated the recent convergence by making key assumptions transparent and intelligible to scholars in many fields. Looking ahead, I discuss recent developments in causal inference that combine the same transparency and relevance.
Econometrica Causal Inference Networks & Graphs · 发表于 2022
Michael P. Leung
摘要
This paper studies causal inference in randomized experiments under network interference. Commonly used models of interference posit that treatments assigned to alters beyond a certain network distance from the ego have no effect on the ego's response. However, this assumption is violated in common models of social interactions. We propose a substantially weaker model of “approximate neighborhood interference” (ANI) under which treatments assigned to alters further from the ego have a smaller, but potentially nonzero, effect on the ego's response. We formally verify that ANI holds for well‐known models of social interactions. Under ANI, restrictions on the network topology, and asymptotics under which the network size increases, we prove that standard inverse‐probability weighting estimators consistently estimate useful exposure effects and are approximately normal. For inference, we consider a network HAC variance estimator. Under a finite population model, we show that the estimator is biased but that the bias can be interpreted as the variance of unit‐level exposure effects. This generalizes Neyman's well‐known result on conservative variance estimation to settings with interference.
JMLR Causal Inference · 发表于 2022
Carlos Fernández-Loría, Foster Provost
摘要
The goal of causal classification is to identify individuals whose outcome would be positively changed by a treatment. Examples include targeting advertisements and targeting retention incentives to reduce churn. Causal classification is challenging because we observe individuals under only one condition (treated or untreated), so we do not know who was influenced by the treatment, but we may estimate the potential outcomes under each condition to decide whom to treat by estimating treatment effects. Curiously, we often see practitioners using simple outcome prediction instead, for example, predicting if someone will purchase if shown the ad. Rather than disregarding this as naive behavior, we present a theoretical analysis comparing treatment effect estimation and outcome prediction when addressing causal classification. We focus on the key question: "When (if ever) is simple outcome prediction preferable to treatment effect estimation for causal classification?" The analysis reveals a causal bias--variance tradeoff. First, when the treatment effect estimation depends on two outcome predictions, larger sampling variance may lead to more errors than the (biased) outcome prediction approach. Second, a stronger signal-to-noise ratio in outcome prediction implies that the bias can help with intervention decisions when outcomes are informative of effects. The theoretical results, as well as simulations, illustrate settings where outcome prediction should actually be better, including cases where (1) the bias may be partially corrected by choosing a different threshold, (2) outcomes and treatment effects are correlated, and (3) data to estimate counterfactuals are limited. A major practical implication is that, for some applications, it might be feasible to make good intervention decisions without any data on how individuals actually behave when intervened. Finally, we show that for a real online advertising application, outcome prediction models indeed excel at causal classification.
JMLR Causal Inference · 发表于 2022
Jaime Roquero Gimenez, Dominik Rothenhäusler
摘要
In causal inference, it is common to estimate the causal effect of a single treatment variable on an outcome. However, practitioners may also be interested in the effect of simultaneous interventions on multiple covariates of a fixed target variable. We propose a novel method that allows to estimate the effect of joint interventions using data from different experiments in which only very few variables are manipulated. If there is only little randomized data or no randomized data at all, one can use observational data sets if certain parental sets are known or instrumental variables are available. If the joint causal effect is linear, the proposed method can be used for estimation and inference of joint causal effects, and we characterize conditions for identifiability. In the overidentified case, we indicate how to leverage all the available causal information across multiple data sets to efficiently estimate the causal effects. If the dimension of the covariate vector is large, we may only have a few samples in each data set. Under a sparsity assumption, we derive an estimator of the causal effects in this high-dimensional scenario. In addition, we show how to deal with the case where a lack of experimental constraints prevents direct estimation of the causal effects. When the joint causal effects are non-linear, we characterize conditions under which identifiability holds, and propose a non-linear causal aggregation methodology for experimental data sets similar to the gradient boosting algorithm where in each iteration we combine weak learners trained on different datasets using only unconfounded samples. We demonstrate the effectiveness of the proposed method on simulated and semi-synthetic data.
JMLR Latent Variables & Mixtures Bayesian Methods · 发表于 2022
Linh Tran, Maja Pantic, Marc Peter Deisenroth
摘要
Recent work in unsupervised learning has focused on efficient inference and learning in latent variables models. Training these models by maximizing the evidence (marginal likelihood) is typically intractable. Thus, a common approximation is to maximize the Evidence Lower BOund (ELBO) instead. Variational autoencoders (VAE) are a powerful and widely-used class of generative models that optimize the ELBO efficiently for large datasets. However, the VAE's default Gaussian choice for the prior imposes a strong constraint on its ability to represent the true posterior, thereby degrading overall performance. A Gaussian mixture model (GMM) would be a richer prior but cannot be handled efficiently within the VAE framework because of the intractability of the Kullback-Leibler divergence for GMMs. We deviate from the common VAE framework in favor of one with an analytical solution for Gaussian mixture prior. To perform efficient inference for GMM priors, we introduce a new constrained objective based on the Cauchy-Schwarz divergence, which can be computed analytically for GMMs. This new objective allows us to incorporate richer, multi-modal priors into the autoencoding framework. We provide empirical studies on a range of datasets and show that our objective improves upon variational auto-encoding models in density estimation, unsupervised clustering, semi-supervised learning, and face analysis.
JMLR Machine Learning · 发表于 2022
Jonathan Ho, Chitwan Saharia, William Chan, David J. Fleet, Mohammad Norouzi, Tim Salimans
摘要
We show that cascaded diffusion models are capable of generating high fidelity images on the class-conditional ImageNet generation benchmark, without any assistance from auxiliary image classifiers to boost sample quality. A cascaded diffusion model comprises a pipeline of multiple diffusion models that generate images of increasing resolution, beginning with a standard diffusion model at the lowest resolution, followed by one or more super-resolution diffusion models that successively upsample the image and add higher resolution details. We find that the sample quality of a cascading pipeline relies crucially on conditioning augmentation, our proposed method of data augmentation of the lower resolution conditioning inputs to the super-resolution models. Our experiments show that conditioning augmentation prevents compounding error during sampling in a cascaded model, helping us to train cascading pipelines achieving FID scores of 1.48 at 64x64, 3.52 at 128x128 and 4.88 at 256x256 resolutions, outperforming BigGAN-deep, and classification accuracy scores of 63.02% (top-1) and 84.06% (top-5) at 256x256, outperforming VQ-VAE-2.
JMLR Testing & Uncertainty High-Dimensional · 发表于 2022
Rafael Izbicki, Gilson Shimizu, Rafael B. Stern
摘要
Conformal methods create prediction bands that control average coverage assuming solely i.i.d. data. Although the literature has mostly focused on prediction intervals, more general regions can often better represent uncertainty. For instance, a bimodal target is better represented by the union of two intervals. Such prediction regions are obtained by CD-split, which combines the split method and a data-driven partition of the feature space which scales to high dimensions. CD-split however contains many tuning parameters, and their role is not clear. In this paper, we provide new insights on CD-split by exploring its theoretical properties. In particular, we show that CD-split converges asymptotically to the oracle highest predictive density set and satisfies local and asymptotic conditional validity. We also present simulations that show how to tune CD-split. Finally, we introduce HPD-split, a variation of CD-split that requires less tuning, and show that it shares the same theoretical guarantees as CD-split. In a wide variety of our simulations, CD-split and HPD-split have better conditional coverage and yield smaller prediction regions than other methods.
Econometrica Econometrics & Economics Causal Inference · 发表于 2022
Atila Abdulkadiroglu, Joshua D. Angrist, Yusuke Narita, Parag Pathak
摘要
Many schools in large urban districts have more applicants than seats. Centralized school assignment algorithms ration seats at over‐subscribed schools using randomly assigned lottery numbers, non‐lottery tie‐breakers like test scores, or both. The New York City public high school match illustrates the latter, using test scores and other criteria to rank applicants at the city's screened schools, combined with lottery tie‐breaking at the rest. We show how to identify causal effects of school attendance in such settings. Our approach generalizes regression discontinuity methods to allow for multiple treatments and multiple running variables, some of which are randomly assigned. The key to this generalization is a local propensity score that quantifies the school assignment probabilities induced by lottery and non‐lottery tie‐breakers. The utility of the local propensity score is demonstrated in an assessment of the predictive value of New York City's school report cards. Schools that earn the highest report card grade indeed improve SAT math scores and increase graduation rates, though by much less than OLS estimates suggest. Selection bias in OLS estimates of grade effects is egregious for screened schools.
JMLR Other · 发表于 2022
Arnak S. Dalalyan, Avetik Karagulyan, Lionel Riou-Durand
摘要
In this paper, we provide non-asymptotic upper bounds on the error of sampling from a target density over $\mathbb{R}^p$ using three schemes of discretized Langevin diffusions. The first scheme is the Langevin Monte Carlo (LMC) algorithm, the Euler discretization of the Langevin diffusion. The second and the third schemes are, respectively, the kinetic Langevin Monte Carlo (KLMC) for differentiable potentials and the kinetic Langevin Monte Carlo for twice-differentiable potentials (KLMC2). The main focus is on the target densities that are smooth and log-concave on $\mathbb{R}^p$, but not necessarily strongly log-concave. Bounds on the computational complexity are obtained under two types of smoothness assumption: the potential has a Lipschitz-continuous gradient and the potential has a Lipschitz-continuous Hessian matrix. The error of sampling is measured by Wasserstein-$q$ distances. We advocate for the use of a new dimension-adapted scaling in the definition of the computational complexity, when Wasserstein-$q$ distances are considered. The obtained results show that the number of iterations to achieve a scaled-error smaller than a prescribed value depends only polynomially in the dimension.
JMLR Optimization & Learning Theory · 发表于 2022
Yichen Zhou, Giles Hooker
摘要
This paper examines a novel gradient boosting framework for regression. We regularize gradient boosted trees by introducing subsampling and employ a modified shrinkage algorithm so that at every boosting stage the estimate is given by an average of trees. The resulting algorithm, titled "Boulevard'", is shown to converge as the number of trees grows. This construction allows us to demonstrate a central limit theorem for this limit, providing a characterization of uncertainty for predictions. A simulation study and real world examples provide support for both the predictive accuracy of the model and its limiting behavior.
JMLR Optimization & Learning Theory · 发表于 2022
Wanrong Zhu, Zhipeng Lou, Wei Biao Wu
摘要
In this paper, we study the concentration property of stochastic gradient descent (SGD) solutions. In existing concentration analyses, researchers impose restrictive requirements on the gradient noise, such as boundedness or sub-Gaussianity. We consider a much richer class of noise where only finitely-many moments are required, thus allowing heavy-tailed noises. In particular, we obtain Nagaev type high-probability upper bounds for the estimation errors of averaged stochastic gradient descent (ASGD) in a linear model. Specifically, we prove that, after $T$ steps of SGD, the ASGD estimate achieves an $O(\sqrt{\log(1/\delta)/T} + (\delta T^{q-1})^{-1/q})$ error rate with probability at least $1-\delta$, where $q>2$ controls the tail of the gradient noise. In comparison, one has the $O(\sqrt{\log(1/\delta)/T})$ error rate for sub-Gaussian noises. We also show that the Nagaev type upper bound is almost tight through an example, where the exact asymptotic form of the tail probability can be derived. Our concentration analysis indicates that, in the case of heavy-tailed noises, the polynomial dependence on the failure probability $\delta$ is generally unavoidable for the error rate of SGD.
Econometrica Econometrics & Economics · 发表于 2022
Manasi Deshpande, Lee M. Lockwood
摘要
The public debate over disability insurance has centered on concerns about individuals without severe health conditions receiving benefits. We go beyond health risk alone to quantify the overall insurance value of U.S. disability programs, including value from insuring nonhealth risk. We find that disability recipients, especially those with less‐severe health conditions, are much more likely to have experienced a wide variety of nonhealth shocks than nonrecipients. Selection into disability receipt on the basis of nonhealth shocks is so strong among individuals with less‐severe health conditions that by many measures less‐severe recipients are worse off than more‐severe recipients. As a result, under baseline assumptions, benefits to less‐severe recipients have an annual surplus value (insurance benefit less efficiency cost) over cost‐equivalent tax cuts of $7700 per recipient, about three‐fourths that of benefits to more‐severe recipients ($9900). Insurance against nonhealth risk accounts for about one‐half of the value of U.S. disability programs.
JMLR Machine Learning Latent Variables & Mixtures · 发表于 2022
Dhruva Tirumala, Alexandre Galashov, Hyeonwoo Noh, Leonard Hasenclever, Razvan Pascanu, Jonathan Schwarz, Guillaume Desjardins, Wojciech Marian Czarnecki, Arun Ahuja, Yee Whye Teh, Nicolas Heess
摘要
As we deploy reinforcement learning agents to solve increasingly challenging problems, methods that allow us to inject prior knowledge about the structure of the world and effective solution strategies becomes increasingly important. In this work we consider how information and architectural constraints can be combined with ideas from the probabilistic modeling literature to learn behavior priors that capture the common movement and interaction patterns that are shared across a set of related tasks or contexts. For example the day-to day behavior of humans comprises distinctive locomotion and manipulation patterns that recur across many different situations and goals. We discuss how such behavior patterns can be captured using probabilistic trajectory models and how these can be integrated effectively into reinforcement learning schemes, e.g. to facilitate multi-task and transfer learning. We then extend these ideas to latent variable models and consider a formulation to learn hierarchical priors that capture different aspects of the behavior in reusable modules. We discuss how such latent variable formulations connect to related work on hierarchical reinforcement learning (HRL) and mutual information and curiosity based objectives, thereby offering an alternative perspective on existing ideas. We demonstrate the effectiveness of our framework by applying it to a range of simulated continuous control domains, videos of which can be found at the following url: https://sites.google.com/view/behavior-priors.
JMLR Bayesian Methods · 发表于 2022
Daniel R. Kowal
摘要
Subset selection is a valuable tool for interpretable learning, scientific discovery, and data compression. However, classical subset selection is often avoided due to selection instability, lack of regularization, and difficulties with post-selection inference. We address these challenges from a Bayesian perspective. Given any Bayesian predictive model M, we extract a family of near-optimal subsets of variables for linear prediction or classification. This strategy deemphasizes the role of a single “best” subset and instead advances the broader perspective that often many subsets are highly competitive. The acceptable family of subsets offers a new pathway for model interpretation and is neatly summarized by key members such as the smallest acceptable subset, along with new (co-) variable importance metrics based on whether variables (co-) appear in all, some, or no acceptable subsets. More broadly, we apply Bayesian decision analysis to derive the optimal linear coefficients for any subset of variables. These coefficients inherit both regularization and predictive uncertainty quantification via M. For both simulated and real data, the proposed approach exhibits better prediction, interval estimation, and variable selection than competing Bayesian and frequentist selection methods. These tools are applied to a large education dataset with highly correlated covariates. Our analysis provides unique insights into the combination of environmental, socioeconomic, and demographic factors that predict educational outcomes, and identifies over 200 distinct subsets of variables that offer near-optimal out-of-sample predictive accuracy.
JMLR Bayesian Methods · 发表于 2022
Terrance D. Savitsky, Matthew R.Williams, Jingchen Hu
摘要
We propose a Bayesian pseudo posterior mechanism to generate record-level synthetic databases equipped with an $(\epsilon,\pi)-$ probabilistic differential privacy (pDP) guarantee, where $\pi$ denotes the probability that any observed database exceeds $\epsilon$. The pseudo posterior mechanism employs a data record-indexed, risk-based weight vector with weight values $\in [0, 1]$ that surgically downweight the likelihood contributions for high-risk records for model estimation and the generation of record-level synthetic data for public release. The pseudo posterior synthesizer constructs a weight for each datum record by using the Lipschitz bound for that record under a log-pseudo likelihood utility function that generalizes the exponential mechanism (EM) used to construct a formally private data generating mechanism. By selecting weights to remove likelihood contributions with non-finite log-likelihood values, we guarantee a finite local privacy guarantee for our pseudo posterior mechanism at every sample size. Our results may be applied to any synthesizing model envisioned by the data disseminator in a computationally tractable way that only involves estimation of a pseudo posterior distribution for parameters, $\theta$, unlike recent approaches that use naturally-bounded utility functions implemented through the EM. We specify conditions that guarantee the asymptotic contraction of $\pi$ to $0$ over the space of databases, such that the form of the guarantee provided by our method is asymptotic. We illustrate our pseudo posterior mechanism on the sensitive family income variable from the Consumer Expenditure Surveys database published by the U.S. Bureau of Labor Statistics. We show that utility is better preserved in the synthetic data for our pseudo posterior mechanism as compared to the EM, both estimated using the same non-private synthesizer, due to our use of targeted downweighting.
JMLR Bayesian Methods Latent Variables & Mixtures · 发表于 2022
Justin D. Silverman, Kimberly Roche, Zachary C. Holmes, Lawrence A. David, Sayan Mukherjee
摘要
Bayesian multinomial logistic-normal (MLN) models are popular for the analysis of sequence count data (e.g., microbiome or gene expression data) due to their ability to model multivariate count data with complex covariance structure. However, existing implementations of MLN models are limited to small datasets due to the non-conjugacy of the multinomial and logistic-normal distributions. Motivated by the need to develop efficient inference for Bayesian MLN models, we develop two key ideas. First, we develop the class of Marginally Latent Matrix-T Process (Marginally LTP) models. We demonstrate that many popular MLN models, including those with latent linear, non-linear, and dynamic linear structure are special cases of this class. Second, we develop an efficient inference scheme for Marginally LTP models with specific accelerations for the MLN subclass. Through application to MLN models, we demonstrate that our inference scheme are both highly accurate and often 4-5 orders of magnitude faster than MCMC.
JMLR Bayesian Methods Networks & Graphs High-Dimensional · 发表于 2022
Yang Ni, Francesco C. Stingo, Veerabhadran Baladandayuthapani
摘要
We introduce Bayesian Gaussian graphical models with covariates (GGMx), a class of multivariate Gaussian distributions with covariate-dependent sparse precision matrix. We propose a general construction of a functional mapping from the covariate space to the cone of sparse positive definite matrices, which encompasses many existing graphical models for heterogeneous settings. Our methodology is based on a novel mixture prior for precision matrices with a non-local component that admits attractive theoretical and empirical properties. The flexible formulation of GGMx allows both the strength and the sparsity pattern of the precision matrix (hence the graph structure) change with the covariates. Posterior inference is carried out with a carefully designed Markov chain Monte Carlo algorithm, which ensures the positive definiteness of sparse precision matrices at any given covariates' values. Extensive simulations and a case study in cancer genomics demonstrate the utility of the proposed model.
JMLR Machine Learning Networks & Graphs · 发表于 2022
Susanna Lange, Kyle Helfrich, Qiang Ye
摘要
Batch normalization (BN) is a popular and ubiquitous method in deep learning that has been shown to decrease training time and improve generalization performance of neural networks. Despite its success, BN is not theoretically well understood. It is not suitable for use with very small mini-batch sizes or online learning. In this paper, we propose a new method called Batch Normalization Preconditioning (BNP). Instead of applying normalization explicitly through a batch normalization layer as is done in BN, BNP applies normalization by conditioning the parameter gradients directly during training. This is designed to improve the Hessian matrix of the loss function and hence convergence during training. One benefit is that BNP is not constrained on the mini-batch size and works in the online learning setting. Furthermore, its connection to BN provides theoretical insights on how BN improves training and how BN is applied to special architectures such as convolutional neural networks. For a theoretical foundation, we also present a novel Hessian condition number based convergence theory for a locally convex but not strong-convex loss, which is applicable to networks with a scale-invariant property.
Econometrica Econometrics & Economics · 发表于 2022
Javier Bianchi, Saki Bigio
摘要
We develop a tractable model of banks' liquidity management with an over‐the‐counter interbank market to study the credit channel of monetary policy. Deposits circulate randomly across banks and must be settled with reserves. We show how monetary policy affects the banking system by altering the trade‐off between profiting from lending and incurring greater liquidity risk. We present two applications of the theory, one involving the connection between the implementation of monetary policy and the pass‐through to lending rates, and another considering a quantitative decomposition behind the collapse in bank lending during the 2008 financial crisis. Our analysis underscores the importance of liquidity frictions and the functioning of interbank markets for the conduct of monetary policy.
JMLR Spatial & Time Series Optimization & Learning Theory · 发表于 2022
Hicham Janati, Marco Cuturi, Alexandre Gramfort
摘要
Several fields in science, from genomics to neuroimaging, require monitoring populations (measures) that evolve with time. These complex datasets, describing dynamics with both time and spatial components, pose new challenges for data analysis. We propose in this work a new framework to carry out averaging of these datasets, with the goal of synthesizing a representative template trajectory from multiple trajectories. We show that this requires addressing three sources of invariance: shifts in time, space, and total population size (or mass/amplitude). Here we draw inspiration from dynamic time warping (DTW), optimal transport (OT) theory and its unbalanced extension (UOT) to propose a criterion that can address all three issues. This proposal leverages a smooth formulation of DTW (Soft-DTW) that is shown to capture temporal shifts, and UOT to handle both variations in space and size. Our proposed loss can be used to define spatio-temporal barycenters as Fréchet means. Using Fenchel duality, we show how these barycenters can be computed efficiently, in parallel, via a novel variant of entropy-regularized debiased UOT. Experiments on handwritten letters and brain imaging data confirm our theoretical findings and illustrate the effectiveness of the proposed loss for spatio-temporal data.
Econometrica Machine Learning Causal Inference · 发表于 2022
Victor Chernozhukov, Whitney K. Newey, Rahul Singh
摘要
Many causal and structural effects depend on regressions. Examples include policy effects, average derivatives, regression decompositions, average treatment effects, causal mediation, and parameters of economic structural models. The regressions may be high‐dimensional, making machine learning useful. Plugging machine learners into identifying equations can lead to poor inference due to bias from regularization and/or model selection. This paper gives automatic debiasing for linear and nonlinear functions of regressions. The debiasing is automatic in using Lasso and the function of interest without the full form of the bias correction. The debiasing can be applied to any regression learner, including neural nets, random forests, Lasso, boosting, and other high‐dimensional methods. In addition to providing the bias correction, we give standard errors that are robust to misspecification, convergence rates for the bias correction, and primitive conditions for asymptotic inference for estimators of a variety of estimators of structural and causal effects. The automatic debiased machine learning is used to estimate the average treatment effect on the treated for the NSW job training data and to estimate demand elasticities from Nielsen scanner data while allowing preferences to be correlated with prices and income.
JMLR Machine Learning · 发表于 2022
Matthias Feurer, Katharina Eggensperger, Stefan Falkner, Marius Lindauer, Frank Hutter
摘要
Automated Machine Learning (AutoML) supports practitioners and researchers with the tedious task of designing machine learning pipelines and has recently achieved substantial success. In this paper, we introduce new AutoML approaches motivated by our winning submission to the second ChaLearn AutoML challenge. We develop PoSH Auto-sklearn, which enables AutoML systems to work well on large datasets under rigid time limits by using a new, simple and meta-feature-free meta-learning technique and by employing a successful bandit strategy for budget allocation. However, PoSH Auto-sklearn introduces even more ways of running AutoML and might make it harder for users to set it up correctly. Therefore, we also go one step further and study the design space of AutoML itself, proposing a solution towards truly hands-free AutoML. Together, these changes give rise to the next generation of our AutoML system, Auto-sklearn 2.0 . We verify the improvements by these additions in an extensive experimental study on 39 AutoML benchmark datasets. We conclude the paper by comparing to other popular AutoML frameworks and Auto-sklearn 1.0 , reducing the relative error by up to a factor of 4.5, and yielding a performance in 10 minutes that is substantially better than what Auto-sklearn 1.0 achieves within an hour.
JMLR Other · 发表于 2022
Jan Niklas Böhm, Philipp Berens, Dmitry Kobak
摘要
Neighbor embeddings are a family of methods for visualizing complex high-dimensional data sets using kNN graphs. To find the low-dimensional embedding, these algorithms combine an attractive force between neighboring pairs of points with a repulsive force between all points. One of the most popular examples of such algorithms is t-SNE. Here we empirically show that changing the balance between the attractive and the repulsive forces in t-SNE using the exaggeration parameter yields a spectrum of embeddings, which is characterized by a simple trade-off: stronger attraction can better represent continuous manifold structures, while stronger repulsion can better represent discrete cluster structures and yields higher kNN recall. We find that UMAP embeddings correspond to t-SNE with increased attraction; mathematical analysis shows that this is because the negative sampling optimization strategy employed by UMAP strongly lowers the effective repulsion. Likewise, ForceAtlas2, commonly used for visualizing developmental single-cell transcriptomic data, yields embeddings corresponding to t-SNE with the attraction increased even more. At the extreme of this spectrum lie Laplacian eigenmaps. Our results demonstrate that many prominent neighbor embedding algorithms can be placed onto the attraction-repulsion spectrum, and highlight the inherent trade-offs between them.
JMLR Other · 发表于 2022
Sébastien Gadat, Ioana Gavra
摘要
This paper studies some asymptotic properties of adaptive algorithms widely used in optimization and machine learning, and among them Adagrad and Rmsprop, which are involved in most of the blackbox deep learning algorithms. Our setup is the non-convex landscape optimization point of view, we consider a one time scale parametrization and the situation where these algorithms may or may not be used with mini-batches. We adopt the point of view of stochastic algorithms and establish the almost sure convergence of these methods when using a decreasing step-size towards the set of critical points of the target function. With a mild extra assumption on the noise, we also obtain the convergence towards the set of minimizers of the function. Along our study, we also obtain a "convergence rate” of the methods, namely a bound on the expected value of the gradient of the cost function along a finite number of iterations.
JMLR Networks & Graphs Meta-Analysis & Data Integration · 发表于 2022
Alex Olshevsky
摘要
We consider whether distributed subgradient methods can achieve a linear speedup over a centralized subgradient method. While it might be hoped that distributed network of $n$ nodes that can compute $n$ times more subgradients in parallel compared to a single node might, as a result, be $n$ times faster, existing bounds for distributed optimization methods are often consistent with a slowdown rather than speedup compared to a single node. We show that a distributed subgradient method has this “linear speedup” property when using a class of square-summable-but-not-summable step-sizes which include $1/t^{\beta}$ when $\beta \in (1/2,1)$; for such step-sizes, we show that after a transient period whose size depends on the spectral gap of the network, the method achieves a performance guarantee that does not depend on the network or the number of nodes. We also show that the same method can fail to have this “asymptotic network independence” property under the optimally decaying step-size $1/\sqrt{t}$ and, as a consequence, can fail to provide a linear speedup compared to a single node with $1/\sqrt{t}$ step-size.
JMLR Other · 发表于 2022
Ping Ma, Yongkai Chen, Xinlian Zhang, Xin Xing, Jingyi Ma, Michael W. Mahoney
摘要
The statistical analysis of Randomized Numerical Linear Algebra (RandNLA) algorithms within the past few years has mostly focused on their performance as point estimators. However, this is insufficient for conducting statistical inference, e.g., constructing confidence intervals and hypothesis testing, since the distribution of the estimator is lacking. In this article, we develop an asymptotic analysis to derive the distribution of RandNLA sampling estimators for the least-squares problem. In particular, we derive the asymptotic distribution of a general sampling estimator with arbitrary sampling probabilities in a fixed design setting. The analysis is conducted in two complementary settings, i.e., when the objective of interest is to approximate the full sample estimator, and when it is to infer the underlying ground truth model parameters. For each setting, we show that the sampling estimator is asymptotically normally distributed under mild regularity conditions. Moreover, the sampling estimator is asymptotically unbiased in both settings. Based on our asymptotic analysis, we use two criteria, the Asymptotic Mean Squared Error (AMSE) and the Expected Asymptotic Mean Squared Error (EAMSE), to identify optimal sampling probabilities. Several of these optimal sampling probability distributions are new to the literature, e.g., the root leverage sampling estimator and the predictor length sampling estimator. Our theoretical results clarify the role of leverage in the sampling process, and our empirical results demonstrate improvements over existing methods.
Econometrica Econometrics & Economics · 发表于 2022
Abe Dunn, Anne Hall, Seidu Dauda
摘要
More than two decades ago, a well‐known study on heart attack treatments provided evidence suggesting that, when appropriately adjusted for quality, medical care prices were actually declining (Cutler, McClellan, Newhouse, and Remler (1998)). Our paper revisits this subject by leveraging estimates from more than 8000 cost‐effectiveness studies across a broad range of conditions and treatments. We find large quality‐adjusted price declines associated with treatment innovations. To incorporate these quality‐adjusted indexes into an aggregate measure of inflation, we combine an unadjusted medical‐care price index, quality‐adjusted price indexes from treatment innovations, and proxies for the diffusion rate of new technologies. In contrast to official statistics that suggest medical care prices increased by 0.53 percent per year relative to economy‐wide inflation from 2000 to 2017, we find that quality‐adjusted medical care prices declined by 1.33 percent per year over the same period.
JMLR Networks & Graphs · 发表于 2022
Chiyuan Zhang, Samy Bengio, Yoram Singer
摘要
Understanding deep neural networks is a major research objective with notable experimental and theoretical attention in recent years. The practical success of excessively large networks underscores the need for better theoretical analyses and justifications. In this paper we focus on layer-wise functional structure and behavior in overparameterized deep models. To do so, we study empirically the layers' robustness to post-training re-initialization and re-randomization of the parameters. We provide experimental results which give evidence for the heterogeneity of layers. Morally, layers of large deep neural networks can be categorized as either "robust" or "critical". Resetting the robust layers to their initial values does not result in adverse decline in performance. In many cases, robust layers hardly change throughout training. In contrast, re-initializing critical layers vastly degrades the performance of the network with test error essentially dropping to random guesses. Our study provides further evidence that mere parameter counting or norm calculations are too coarse in studying generalization of deep models, and "flatness" and robustness analysis of trained models need to be examined while taking into account the respective network architectures.
JMLR Optimization & Learning Theory Machine Learning Networks & Graphs · 发表于 2022
Zhong Li, Jiequn Han, Weinan E, Qianxiao Li
摘要
We perform a systematic study of the approximation properties and optimization dynamics of recurrent neural networks (RNNs) when applied to learn input-output relationships in temporal data. We consider the simple but representative setting of using continuous-time linear RNNs to learn from data generated by linear relationships. On the approximation side, we prove a direct and an inverse approximation theorem of linear functionals using RNNs, which reveal the intricate connections between memory structures in the target and the corresponding approximation efficiency. In particular, we show that temporal relationships can be effectively approximated by RNNs if and only if the former possesses sufficient memory decay. On the optimization front, we perform detailed analysis of the optimization dynamics, including a precise understanding of the difficulty that may arise in learning relationships with long-term memory. The term “curse of memory” is coined to describe the uncovered phenomena, akin to the “curse of dimension” that plagues high-dimensional function approximation. These results form a relatively complete picture of the interaction of memory and recurrent structures in the linear dynamical setting.
JMLR Machine Learning · 发表于 2022
Jayakumar Subramanian, Amit Sinha, Raihan Seraj, Aditya Mahajan
摘要
We propose a theoretical framework for approximate planning and learning in partially observed systems. Our framework is based on the fundamental notion of information state. We provide two definitions of information state---i) a function of history which is sufficient to compute the expected reward and predict its next value; ii) a function of the history which can be recursively updated and is sufficient to compute the expected reward and predict the next observation. An information state always leads to a dynamic programming decomposition. Our key result is to show that if a function of the history (called AIS) approximately satisfies the properties of the information state, then there is a corresponding approximate dynamic program. We show that the policy computed using this is approximately optimal with bounded loss of optimality. We show that several approximations in state, observation and action spaces in literature can be viewed as instances of AIS. In some of these cases, we obtain tighter bounds. A salient feature of AIS is that it can be learnt from data. We present AIS based multi-time scale policy gradient algorithms and detailed numerical experiments with low, moderate and high dimensional environments.
JMLR Bayesian Methods Nonparametric & Semiparametric · 发表于 2022
Yuexi Wang, Tetsuya Kaji, Veronika Rockova
摘要
Approximate Bayesian Computation (ABC) enables statistical inference in simulator-based models whose likelihoods are difficult to calculate but easy to simulate from. ABC constructs a kernel-type approximation to the posterior distribution through an accept/reject mechanism which compares summary statistics of real and simulated data. To obviate the need for summary statistics, we directly compare empirical distributions with a Kullback-Leibler (KL) divergence estimator obtained via contrastive learning. In particular, we blend flexible machine learning classifiers within ABC to automate fake/real data comparisons. We consider the traditional accept/reject kernel as well as an exponential weighting scheme which does not require the ABC acceptance threshold. Our theoretical results show that the rate at which our ABC posterior distributions concentrate around the true parameter depends on the estimation error of the classifier. We derive limiting posterior shape results and find that, with a properly scaled exponential kernel, asymptotic normality holds. We demonstrate the usefulness of our approach on simulated examples as well as real data in the context of stock volatility estimation.
JMLR Machine Learning Networks & Graphs Bayesian Methods · 发表于 2022
Luong-Ha Nguyen, James-A. Goulet
摘要
With few exceptions, neural networks have been relying on backpropagation and gradient descent as the inference engine in order to learn the model parameters, because closed-form Bayesian inference for neural networks has been considered to be intractable. In this paper, we show how we can leverage the tractable approximate Gaussian inference's (TAGI) capabilities to infer hidden states, rather than only using it for inferring the network's parameters. One novel aspect is that it allows inferring hidden states through the imposition of constraints designed to achieve specific objectives, as illustrated through three examples: (1) the generation of adversarial-attack examples, (2) the usage of a neural network as a black-box optimization method, and (3) the application of inference on continuous-action reinforcement learning. In these three examples, the constrains are in (1), a target label chosen to fool a neural network, and in (2 and 3) the derivative of the network with respect to its input that is set to zero in order to infer the optimal input values that are either maximizing or minimizing it. These applications showcase how tasks that were previously reserved to gradient-based optimization approaches can now be approached with analytically tractable inference.
JMLR Bayesian Methods Optimization & Learning Theory · 发表于 2022
Jeremias Knoblauch, Jack Jewson, Theodoros Damoulas
摘要
We advocate an optimization-centric view of Bayesian inference. Our inspiration is the representation of Bayes' rule as infinite-dimensional optimization (Csiszar, 1975; Donsker and Varadhan, 1975; Zellner, 1988). Equipped with this perspective, we study Bayesian inference when one does not have access to (1) well-specified priors, (2) well-specified likelihoods, (3) infinite computing power. While these three assumptions underlie the standard Bayesian paradigm, they are typically inappropriate for modern Machine Learning applications. We propose addressing this through an optimization-centric generalization of Bayesian posteriors that we call the Rule of Three (RoT). The RoT can be justified axiomatically and recovers Bayesian, PAC-Bayesian and VI posteriors as special cases. While the RoT is primarily a conceptual and theoretical device, it also encompasses a novel sub-class of tractable posteriors which we call Generalized Variational Inference (GVI) posteriors. Just as the RoT, GVI posteriors are specified by three arguments: a loss, a divergence and a variational family. They also possess a number of desirable properties, including modularity, Frequentist consistency and an interpretation as approximate ELBO. We explore applications of GVI posteriors, and show that they can be used to improve robustness and posterior marginals on Bayesian Neural Networks and Deep Gaussian Processes.
JMLR Bayesian Methods · 发表于 2022
Jaouad Mourtada, Stéphane Gaïffas
摘要
We introduce a procedure for conditional density estimation under logarithmic loss, which we call SMP (Sample Minmax Predictor). This estimator minimizes a new general excess risk bound for statistical learning. On standard examples, this bound scales as $d/n$ with $d$ the model dimension and $n$ the sample size, and critically remains valid under model misspecification. Being an improper (out-of-model) procedure, SMP improves over within-model estimators such as the maximum likelihood estimator, whose excess risk degrades under misspecification. Compared to approaches reducing to the sequential problem, our bounds remove suboptimal $\log n$ factors and can handle unbounded classes. For the Gaussian linear model, the predictions and risk bound of SMP are governed by leverage scores of covariates, nearly matching the optimal risk in the well-specified case without conditions on the noise variance or approximation error of the linear model. For logistic regression, SMP provides a non-Bayesian approach to calibration of probabilistic predictions relying on virtual samples, and can be computed by solving two logistic regressions. It achieves a non-asymptotic excess risk of $O((d + B^2R^2)/n)$, where $R$ bounds the norm of features and $B$ that of the comparison parameter; by contrast, no within-model estimator can achieve better rate than $\min({B R}/{\sqrt{n}}, {d e^{BR}}/{n} )$ in general. This provides a more practical alternative to Bayesian approaches, which require approximate posterior sampling, thereby partly addressing a question raised by Foster et al. (2018).
JMLR Networks & Graphs · 发表于 2022
Jian Huang, Yuling Jiao, Zhen Li, Shiao Liu, Yang Wang, Yunfei Yang
摘要
This paper studies how well generative adversarial networks (GANs) learn probability distributions from finite samples. Our main results establish the convergence rates of GANs under a collection of integral probability metrics defined through Hölder classes, including the Wasserstein distance as a special case. We also show that GANs are able to adaptively learn data distributions with low-dimensional structures or have Hölder densities, when the network architectures are chosen properly. In particular, for distributions concentrated around a low-dimensional set, we show that the learning rates of GANs do not depend on the high ambient dimension, but on the lower intrinsic dimension. Our analysis is based on a new oracle inequality decomposing the estimation error into the generator and discriminator approximation error and the statistical error, which may be of independent interest.
JMLR Other · 发表于 2022
Wenlong Mou, Nicolas Flammarion, Martin J. Wainwright, Peter L. Bartlett
摘要
We consider the problem of sampling from a density of the form $p(x) \propto \exp(-f(x)- g(x))$, where $f: \mathbb{R}^d \rightarrow \mathbb{R}$ is a smooth function and $g: \mathbb{R}^d \rightarrow \mathbb{R}$ is a convex and Lipschitz function. We propose a new algorithm based on the Metropolis--Hastings framework. Under certain isoperimetric inequalities on the target density, we prove that the algorithm mixes to within total variation (TV) distance $\varepsilon$ of the target density in at most $O(d \log (d/\varepsilon))$ iterations. This guarantee extends previous results on sampling from distributions with smooth log densities ($g = 0$) to the more general composite non-smooth case, with the same mixing time up to a multiple of the condition number. Our method is based on a novel proximal-based proposal distribution that can be efficiently computed for a large class of non-smooth functions $g$. Simulation results on posterior sampling problems that arise from the Bayesian Lasso show empirical advantage over previous proposal distributions.
JMLR Bayesian Methods Machine Learning Networks & Graphs · 发表于 2022
Ba-Hien Tran, Simone Rossi, Dimitrios Milios, Maurizio Filippone
摘要
The Bayesian treatment of neural networks dictates that a prior distribution is specified over their weight and bias parameters. This poses a challenge because modern neural networks are characterized by a large number of parameters, and the choice of these priors has an uncontrolled effect on the induced functional prior, which is the distribution of the functions obtained by sampling the parameters from their prior distribution. We argue that this is a hugely limiting aspect of Bayesian deep learning, and this work tackles this limitation in a practical and effective way. Our proposal is to reason in terms of functional priors, which are easier to elicit, and to “tune” the priors of neural network parameters in a way that they reflect such functional priors. Gaussian processes offer a rigorous framework to define prior distributions over functions, and we propose a novel and robust framework to match their prior with the functional prior of neural networks based on the minimization of their Wasserstein distance. We provide vast experimental evidence that coupling these priors with scalable Markov chain Monte Carlo sampling offers systematically large performance improvements over alternative choices of priors and state-of-the-art approximate Bayesian deep learning approaches. We consider this work a considerable step in the direction of making the long-standing challenge of carrying out a fully Bayesian treatment of neural networks, including convolutional neural networks, a concrete possibility.
Econometrica Other · 发表于 2022
Tayfun Sönmez, M. Bumin Yenmez
摘要
Sanctioned by its constitution, India is home to the world's most comprehensive affirmative action program, where historically discriminated groups are protected with vertical reservations implemented as “set asides,” and other disadvantaged groups are protected with horizontal reservations implemented as “minimum guarantees.” A mechanism mandated by the Supreme Court in 1995 suffers from important anomalies, triggering countless litigations in India. Foretelling a recent reform correcting the flawed mechanism, we propose the 2SMG mechanism that resolves all anomalies, and characterize it with desiderata reflecting laws of India. Subsequently rediscovered with a high court judgment and enforced in Gujarat, 2SMG is also endorsed by Saurav Yadav v. State of UP (2020), in a Supreme Court ruling that rescinded the flawed mechanism. While not explicitly enforced, 2SMG is indirectly enforced for an important subclass of applications in India, because no other mechanism satisfies the new mandates of the Supreme Court.