The study of almost surely discrete random probability measures is an active line of research in Bayesian nonparametrics. The idea of assuming interaction across the atoms of the random probability measure has recently spurred significant interest in the context of Bayesian mixture models. This allows the definition of priors that encourage well separated and interpretable clusters. In this work, we provide a unified framework for the construction and the Bayesian analysis of random probability measures with interacting atoms, encompassing both repulsive and attractive behaviors. Specifically we derive closed-form expressions for the posterior distribution, the marginal and predictive distributions, which were not previously available except for the case of measures with i.i.d. atoms. We show how these quantities are fundamental both for prior elicitation and to develop new posterior simulation algorithms for hierarchical mixture models. Our results are obtained without any assumption on the finite point process that governs the atoms of the random measure. Their proofs rely on new analytical tools borrowed from the theory of Palm calculus and that might be of independent interest. We specialize our treatment to the classes of Poisson, Gibbs, and Determinantal point processes, as well as to the case of shot-noise Cox processes.
相關內容
There is a growing interest in using reinforcement learning (RL) to personalize sequences of treatments in digital health to support users in adopting healthier behaviors. Such sequential decision-making problems involve decisions about when to treat and how to treat based on the user's context (e.g., prior activity level, location, etc.). Online RL is a promising data-driven approach for this problem as it learns based on each user's historical responses and uses that knowledge to personalize these decisions. However, to decide whether the RL algorithm should be included in an ``optimized'' intervention for real-world deployment, we must assess the data evidence indicating that the RL algorithm is actually personalizing the treatments to its users. Due to the stochasticity in the RL algorithm, one may get a false impression that it is learning in certain states and using this learning to provide specific treatments. We use a working definition of personalization and introduce a resampling-based methodology for investigating whether the personalization exhibited by the RL algorithm is an artifact of the RL algorithm stochasticity. We illustrate our methodology with a case study by analyzing the data from a physical activity clinical trial called HeartSteps, which included the use of an online RL algorithm. We demonstrate how our approach enhances data-driven truth-in-advertising of algorithm personalization both across all users as well as within specific users in the study.
Hierarchical forecasting problems arise when time series have a natural group structure, and predictions at multiple levels of aggregation and disaggregation across the groups are needed. In such problems, it is often desired to satisfy the aggregation constraints in a given hierarchy, referred to as hierarchical coherence in the literature. Maintaining coherence while producing accurate forecasts can be a challenging problem, especially in the case of probabilistic forecasting. We present a novel method capable of accurate and coherent probabilistic forecasts for time series when reliable hierarchical information is present. We call it Deep Poisson Mixture Network (DPMN). It relies on the combination of neural networks and a statistical model for the joint distribution of the hierarchical multivariate time series structure. By construction, the model guarantees hierarchical coherence and provides simple rules for aggregation and disaggregation of the predictive distributions. We perform an extensive empirical evaluation comparing the DPMN to other state-of-the-art methods which produce hierarchically coherent probabilistic forecasts on multiple public datasets. Comparing to existing coherent probabilistic models, we obtain a relative improvement in the overall Continuous Ranked Probability Score (CRPS) of 11.8% on Australian domestic tourism data, and 8.1% on the Favorita grocery sales dataset, where time series are grouped with geographical hierarchies or travel intent hierarchies. For San Francisco Bay Area highway traffic, where the series' hierarchical structure is randomly assigned, and their correlations are less informative, our method does not show significant performance differences over statistical baselines.
Bayesian nonparametric hierarchical priors provide flexible models for sharing of information within and across groups. We focus on latent feature allocation models, where the data structures correspond to multisets or unbounded sparse matrices. The fundamental development in this regard is the Hierarchical Indian Buffet process (HIBP), devised by Thibaux and Jordan (2007). However, little is known in terms of explicit tractable descriptions of the joint, marginal, posterior and predictive distributions of the HIBP. We provide explicit novel descriptions of these quantities, in the Bernoulli HIBP and general spike and slab HIBP settings, which allows for exact sampling and simpler practical implementation. We then extend these results to the more complex setting of hierarchies of general HIBP (HHIBP). The generality of our framework allows one to recognize important structure that may otherwise be masked in the Bernoulli setting, and involves characterizations via dynamic mixed Poisson random count matrices. Our analysis shows that the standard choice of hierarchical Beta processes for modeling across group sharing is not ideal in the classic Bernoulli HIBP setting proposed by Thibaux and Jordan (2007), or other spike and slab HIBP settings, and we thus indicate tractable alternative priors.
Entity alignment (EA) aims to discover the equivalent entities in different knowledge graphs (KGs), which play an important role in knowledge engineering. Recently, EA with dangling entities has been proposed as a more realistic setting, which assumes that not all entities have corresponding equivalent entities. In this paper, we focus on this setting. Some work has explored this problem by leveraging translation API, pre-trained word embeddings, and other off-the-shelf tools. However, these approaches over-rely on the side information (e.g., entity names), and fail to work when the side information is absent. On the contrary, they still insufficiently exploit the most fundamental graph structure information in KG. To improve the exploitation of the structural information, we propose a novel entity alignment framework called Weakly-Optimal Graph Contrastive Learning (WOGCL), which is refined on three dimensions : (i) Model. We propose a novel Gated Graph Attention Network to capture local and global graph structure similarity. (ii) Training. Two learning objectives: contrastive learning and optimal transport learning are designed to obtain distinguishable entity representations via the optimal transport plan. (iii) Inference. In the inference phase, a PageRank-based method is proposed to calculate higher-order structural similarity. Extensive experiments on two dangling benchmarks demonstrate that our WOGCL outperforms the current state-of-the-art methods with pure structural information in both traditional (relaxed) and dangling (consolidated) settings. The code will be public soon.
The Transformer is a highly successful deep learning model that has revolutionised the world of artificial neural networks, first in natural language processing and later in computer vision. This model is based on the attention mechanism and is able to capture complex semantic relationships between a variety of patterns present in the input data. Precisely because of these characteristics, the Transformer has recently been exploited for time series forecasting problems, assuming its natural adaptability to the domain of continuous numerical series. Despite the acclaimed results in the literature, some works have raised doubts about the robustness of this approach. In this paper, we further investigate the effectiveness of Transformer-based models applied to the domain of time series forecasting, demonstrate their limitations, and propose a set of alternative models that are better performing and significantly less complex. In particular, we empirically show how simplifying this forecasting model almost always leads to an improvement, reaching the state of the art among Transformer-based architectures. We also propose shallow models without the attention mechanism, which compete with the overall state of the art in long time series forecasting, and demonstrate their ability to accurately predict extremely long windows. We show how it is always necessary to use a simple baseline to verify the effectiveness of one's models, and finally we conclude the paper with a reflection on recent research paths and the desire to follow trends and apply the latest model even where it may not be necessary.
In this paper we investigate phenomena of spontaneous emergence or purposeful formation of highly organized structures in networks of related agents. We show that the formation of large organized structures requires exponentially large, in the size of the structures, networks. Our approach is based on Kolmogorov, or descriptional, complexity of networks viewed as finite size strings. We apply this approach to the study of the emergence or formation of simple organized, hierarchical, structures based on Sierpinski Graphs and we prove a Ramsey type theorem that bounds the number of vertices in Kolmogorov random graphs that contain Sierpinski Graphs as subgraphs. Moreover, we show that Sierpinski Graphs encompass close-knit relationships among their vertices that facilitate fast spread and learning of information when agents in their vertices are engaged in pairwise interactions modelled as two person games. Finally, we generalize our findings for any organized structure with succinct representations. Our work can be deployed, in particular, to study problems related to the security of networks by identifying conditions which enable or forbid the formation of sufficiently large insider subnetworks with malicious common goal to overtake the network or cause disruption of its operation.
Deep generative models (DGMs) are data-eager because learning a complex model on limited data suffers from a large variance and easily overfits. Inspired by the classical perspective of the bias-variance tradeoff, we propose regularized deep generative model (Reg-DGM), which leverages a nontransferable pre-trained model to reduce the variance of generative modeling with limited data. Formally, Reg-DGM optimizes a weighted sum of a certain divergence and the expectation of an energy function, where the divergence is between the data and the model distributions, and the energy function is defined by the pre-trained model w.r.t. the model distribution. We analyze a simple yet representative Gaussian-fitting case to demonstrate how the weighting hyperparameter trades off the bias and the variance. Theoretically, we characterize the existence and the uniqueness of the global minimum of Reg-DGM in a non-parametric setting and prove its convergence with neural networks trained by gradient-based methods. Empirically, with various pre-trained feature extractors and a data-dependent energy function, Reg-DGM consistently improves the generation performance of strong DGMs with limited data and achieves competitive results to the state-of-the-art methods. Our implementation is available at //github.com/ML-GSAI/Reg-ADA-APA.
We present an alternating least squares type numerical optimization scheme to estimate conditionally-independent mixture models in $\mathbb{R}^n$, without parameterizing the distributions. Following the method of moments, we tackle an incomplete tensor decomposition problem to learn the mixing weights and componentwise means. Then we compute the cumulative distribution functions, higher moments and other statistics of the component distributions through linear solves. Crucially for computations in high dimensions, the steep costs associated with high-order tensors are evaded, via the development of efficient tensor-free operations. Numerical experiments demonstrate the competitive performance of the algorithm, and its applicability to many models and applications. Furthermore we provide theoretical analyses, establishing identifiability from low-order moments of the mixture and guaranteeing local linear convergence of the ALS algorithm.
Bayesian causal structure learning aims to learn a posterior distribution over directed acyclic graphs (DAGs), and the mechanisms that define the relationship between parent and child variables. By taking a Bayesian approach, it is possible to reason about the uncertainty of the causal model. The notion of modelling the uncertainty over models is particularly crucial for causal structure learning since the model could be unidentifiable when given only a finite amount of observational data. In this paper, we introduce a novel method to jointly learn the structure and mechanisms of the causal model using Variational Bayes, which we call Variational Bayes-DAG-GFlowNet (VBG). We extend the method of Bayesian causal structure learning using GFlowNets to learn not only the posterior distribution over the structure, but also the parameters of a linear-Gaussian model. Our results on simulated data suggest that VBG is competitive against several baselines in modelling the posterior over DAGs and mechanisms, while offering several advantages over existing methods, including the guarantee to sample acyclic graphs, and the flexibility to generalize to non-linear causal mechanisms.
Modern datasets commonly feature both substantial missingness and many variables of mixed data types, which present significant challenges for estimation and inference. Complete case analysis, which proceeds using only the observations with fully-observed variables, is often severely biased, while model-based imputation of missing values is limited by the ability of the model to capture complex dependencies among (possibly many) variables of mixed data types. To address these challenges, we develop a novel Bayesian mixture copula for joint and nonparametric modeling of multivariate count, continuous, ordinal, and unordered categorical variables, and deploy this model for inference, prediction, and imputation of missing data. Most uniquely, we introduce a new and computationally efficient strategy for marginal distribution estimation that eliminates the need to specify any marginal models yet delivers posterior consistency for each marginal distribution and the copula parameters under missingness-at-random. Extensive simulation studies demonstrate exceptional modeling and imputation capabilities relative to competing methods, especially with mixed data types, complex missingness mechanisms, and nonlinear dependencies. We conclude with a data analysis that highlights how improper treatment of missing data can distort a statistical analysis, and how the proposed approach offers a resolution.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191