We propose a very fast approximate Markov Chain Monte Carlo (MCMC) sampling framework that is applicable to a large class of sparse Bayesian inference problems, where the computational cost per iteration in several models is of order $O(ns)$, where $n$ is the sample size, and $s$ the underlying sparsity of the model. This cost can be further reduced by data sub-sampling when stochastic gradient Langevin dynamics are employed. The algorithm is an extension of the asynchronous Gibbs sampler of Johnson et al. (2013), but can be viewed from a statistical perspective as a form of Bayesian iterated sure independent screening (Fan et al. (2009)). We show that in high-dimensional linear regression problems, the Markov chain generated by the proposed algorithm admits an invariant distribution that recovers correctly the main signal with high probability under some statistical assumptions. Furthermore we show that its mixing time is at most linear in the number of regressors. We illustrate the algorithm with several models.
相關內容
Over the last decades, various "non-linear" MCMC methods have arisen. While appealing for their convergence speed and efficiency, their practical implementation and theoretical study remain challenging. In this paper, we introduce a non-linear generalization of the Metropolis-Hastings algorithm to a proposal that depends not only on the current state, but also on its law. We propose to simulate this dynamics as the mean field limit of a system of interacting particles, that can in turn itself be understood as a generalisation of the Metropolis-Hastings algorithm to a population of particles. Under the double limit in number of iterations and number of particles we prove that this algorithm converges. Then, we propose an efficient GPU implementation and illustrate its performance on various examples. The method is particularly stable on multimodal examples and converges faster than the classical methods.
Researchers are often interested in learning not only the effect of treatments on outcomes, but also the pathways through which these effects operate. A mediator is a variable that is affected by treatment and subsequently affects outcome. Existing methods for penalized mediation analyses either assume that finite-dimensional linear models are sufficient to remove confounding bias, or perform no confounding control at all. In practice, these assumptions may not hold. We propose a method that considers the confounding functions as nuisance parameters to be estimated using data-adaptive methods. We then use a novel regularization method applied to this objective function to identify a set of important mediators. We derive the asymptotic properties of our estimator and establish the oracle property under certain assumptions. Asymptotic results are also presented in a local setting which contrast the proposal with the standard adaptive lasso. We also propose a perturbation bootstrap technique to provide asymptotically valid post-selection inference for the mediated effects of interest. The performance of these methods will be discussed and demonstrated through simulation studies.
Faster inference of deep learning models is highly demanded on edge devices and even servers, for both financial and environmental reasons. To address this issue, we propose SoftNeuro, a novel, high-performance inference framework with efficient performance tuning. The key idea is to separate algorithmic routines from network layers. Our framework maximizes the inference performance by profiling various routines for each layer and selecting the fastest path. To efficiently find the best path, we propose a routine-selection algorithm based on dynamic programming. Experiments show that the proposed framework achieves both fast inference and efficient tuning.
Stochastic gradient Markov chain Monte Carlo (SGMCMC) is considered the gold standard for Bayesian inference in large-scale models, such as Bayesian neural networks. Since practitioners face speed versus accuracy tradeoffs in these models, variational inference (VI) is often the preferable option. Unfortunately, VI makes strong assumptions on both the factorization and functional form of the posterior. In this work, we propose a new non-parametric variational approximation that makes no assumptions about the approximate posterior's functional form and allows practitioners to specify the exact dependencies the algorithm should respect or break. The approach relies on a new Langevin-type algorithm that operates on a modified energy function, where parts of the latent variables are averaged over samples from earlier iterations of the Markov chain. This way, statistical dependencies can be broken in a controlled way, allowing the chain to mix faster. This scheme can be further modified in a "dropout" manner, leading to even more scalability. By implementing the scheme on a ResNet-20 architecture, we obtain better predictive likelihoods and larger effective sample sizes than full SGMCMC.
In this paper, we introduce a reversible version of a genetically modified mode jumping Markov chain Monte Carlo algorithm (GMJMCMC) for inference on posterior model probabilities in complex model spaces, where the number of explanatory variables is prohibitively large for classical Markov Chain Monte Carlo methods. Unlike the earlier proposed GMJMCMC algorithm, the introduced algorithm is a proper MCMC and its limiting distribution corresponds to the posterior marginal model probabilities in the explored model space under reasonable regularity conditions.
In this article, we introduce mixture representations for likelihood ratio ordered distributions. Essentially, the ratio of two probability densities, or mass functions, is monotone if and only if one can be expressed as a mixture of one-sided truncations of the other. To illustrate the practical value of the mixture representations, we address the problem of density estimation for likelihood ratio ordered distributions. In particular, we propose a nonparametric Bayesian solution which takes advantage of the mixture representations. The prior distribution is constructed from Dirichlet process mixtures and has large support on the space of pairs of densities satisfying the monotone ratio constraint. With a simple modification to the prior distribution, we can test the equality of two distributions against the alternative of likelihood ratio ordering. We develop a Markov chain Monte Carlo algorithm for posterior inference and demonstrate the method in a biomedical application.
In recent years, the transformer has established itself as a workhorse in many applications ranging from natural language processing to reinforcement learning. Similarly, Bayesian deep learning has become the gold-standard for uncertainty estimation in safety-critical applications, where robustness and calibration are crucial. Surprisingly, no successful attempts to improve transformer models in terms of predictive uncertainty using Bayesian inference exist. In this work, we study this curiously underpopulated area of Bayesian transformers. We find that weight-space inference in transformers does not work well, regardless of the approximate posterior. We also find that the prior is at least partially at fault, but that it is very hard to find well-specified weight priors for these models. We hypothesize that these problems stem from the complexity of obtaining a meaningful mapping from weight-space to function-space distributions in the transformer. Therefore, moving closer to function-space, we propose a novel method based on the implicit reparameterization of the Dirichlet distribution to apply variational inference directly to the attention weights. We find that this proposed method performs competitively with our baselines.
We develop a post-selective Bayesian framework to jointly and consistently estimate parameters in group-sparse linear regression models. After selection with the Group LASSO (or generalized variants such as the overlapping, sparse, or standardized Group LASSO), uncertainty estimates for the selected parameters are unreliable in the absence of adjustments for selection bias. Existing post-selective approaches are limited to uncertainty estimation for (i) real-valued projections onto very specific selected subspaces for the group-sparse problem, (ii) selection events categorized broadly as polyhedral events that are expressible as linear inequalities in the data variables. Our Bayesian methods address these gaps by deriving a likelihood adjustment factor, and an approximation thereof, that eliminates bias from selection. Paying a very nominal price for this adjustment, experiments on simulated data, and data from the Human Connectome Project demonstrate the efficacy of our methods for a joint estimation of group-sparse parameters and their uncertainties post selection.
Inference of directed relations given some unspecified interventions, that is, the target of each intervention is not known, is important yet challenging. For instance, it is of high interest to unravel the regulatory roles of genes with inherited genetic variants like single-nucleotide polymorphisms (SNPs), which can be unspecified interventions because of their regulatory function on some unknown genes. In this article, we test hypothesized directed relations with unspecified interventions. First, we derive conditions to yield an identifiable model. Unlike classical inference, hypothesis testing requires identifying ancestral relations and relevant interventions for each hypothesis-specific primary variable, referring to as causal discovery. Towards this end, we propose a peeling algorithm to establish a hierarchy of primary variables as nodes, starting with leaf nodes at the hierarchy's bottom, for which we derive a difference-of-convex (DC) algorithm for nonconvex minimization. Moreover, we prove that the peeling algorithm yields consistent causal discovery, and the DC algorithm is a low-order polynomial algorithm capable of finding a global minimizer almost surely under the data generating distribution. Second, we propose a modified likelihood ratio test, eliminating nuisance parameters to increase power. To enhance finite-sample performance, we integrate the modified likelihood ratio test with a data perturbation scheme by accounting for the uncertainty of identifying ancestral relations and relevant interventions. Also, we show that the distribution of a data-perturbation test statistic converges to the target distribution in high dimensions. Numerical examples demonstrate the utility and effectiveness of the proposed methods, including an application to infer gene regulatory networks.
Amortized inference has led to efficient approximate inference for large datasets. The quality of posterior inference is largely determined by two factors: a) the ability of the variational distribution to model the true posterior and b) the capacity of the recognition network to generalize inference over all datapoints. We analyze approximate inference in variational autoencoders in terms of these factors. We find that suboptimal inference is often due to amortizing inference rather than the limited complexity of the approximating distribution. We show that this is due partly to the generator learning to accommodate the choice of approximation. Furthermore, we show that the parameters used to increase the expressiveness of the approximation play a role in generalizing inference rather than simply improving the complexity of the approximation.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191