A new family of penalty functions, adaptive to likelihood, is introduced for model selection in general regression models. It arises naturally through assuming certain types of prior distribution on the regression parameters. To study stability properties of the penalized maximum likelihood estimator, two types of asymptotic stability are defined. Theoretical properties, including the parameter estimation consistency, model selection consistency, and asymptotic stability, are established under suitable regularity conditions. An efficient coordinate-descent algorithm is proposed. Simulation results and real data analysis show that the proposed method has competitive performance in comparison with existing ones.

Decision tree learning is a popular approach for classification and regression in machine learning and statistics, and Bayesian formulations---which introduce a prior distribution over decision trees, and formulate learning as posterior inference given data---have been shown to produce competitive performance. Unlike classic decision tree learning algorithms like ID3, C4.5 and CART, which work in a top-down manner, existing Bayesian algorithms produce an approximation to the posterior distribution by evolving a complete tree (or collection thereof) iteratively via local Monte Carlo modifications to the structure of the tree, e.g., using Markov chain Monte Carlo (MCMC). We present a sequential Monte Carlo (SMC) algorithm that instead works in a top-down manner, mimicking the behavior and speed of classic algorithms. We demonstrate empirically that our approach delivers accuracy comparable to the most popular MCMC method, but operates more than an order of magnitude faster, and thus represents a better computation-accuracy tradeoff.

This supplementary material includes three parts: some preliminary results, four examples, an experiment, three new algorithms, and all proofs of the results in the paper "Reversible MCMC on Markov equivalence classes of sparse directed acyclic graphs".

We review the challenges of Bayesian network learning, especially parameter learning, and specify the problem of learning with sparse data. We explain how it is possible to incorporate both qualitative knowledge and data with a multinomial parameter learning method to achieve more accurate predictions with sparse data.

This paper proposes a simple linear Bayesian approach to reinforcement learning. We show that with an appropriate basis, a Bayesian linear Gaussian model is sufficient for accurately estimating the system dynamics, and in particular when we allow for correlated noise. Policies are estimated by first sampling a transition model from the current posterior, and then performing approximate dynamic programming on the sampled model. This form of approximate Thompson sampling results in good exploration in unknown environments. The approach can also be seen as a Bayesian generalisation of least-squares policy iteration, where the empirical transition matrix is replaced with a sample from the posterior.

A novel proposal for object recognition based on relational grammars and Bayesian Networks is presented. Based on this grammar an object is represented as a hierarchy of features and spatial relations. This representation is transformed to a Bayesian network structure which parameters are learned from examples. Thus, recognition is based on probabilistic inference in the Bayesian network representation. Preliminary results in modeling natural objects are presented.

This paper concerns building probabilistic models with an underlying ontology that defines the classes and properties used in the model. In particular, it considers the problem of reasoning with properties that may not always be defined. Furthermore, we may even be uncertain about whether a property is defined for a given individual. One approach is to explicitly add a value "undefined" to the range of random variables, forming extended belief networks; however, adding an extra value to a random variable's range has a large computational overhead. In this paper, we propose an alternative, ontologically-based belief networks, where all properties are only used when they are defined, and we show how probabilistic reasoning can be carried out without explicitly using the value "undefined" during inference. We prove this is equivalent to reasoning with the corresponding extended belief network and empirically demonstrate that inference becomes more efficient.

We present a novel approach for multilabel classification based on an ensemble of Bayesian networks. The class variables are connected by a tree; each model of the ensemble uses a different class as root of the tree. We assume the features to be conditionally independent given the classes, thus generalizing the naive Bayes assumption to the multiclass case. This assumption allows us to optimally identify the correlations between classes and features; such correlations are moreover shared across all models of the ensemble. Inferences are drawn from the ensemble via logarithmic opinion pooling. To minimize Hamming loss, we compute the marginal probability of the classes by running standard inference on each Bayesian network in the ensemble, and then pooling the inferences. To instead minimize the subset 0/1 loss, we pool the joint distributions of each model and cast the problem as a MAP inference in the corresponding graphical model. Experiments show that the approach is competitive with state-of-the-art methods for multilabel classification.

We target the problem of accuracy and robustness in causal inference from finite data sets. Our aim is to combine the inherent robustness of the Bayesian approach with the theoretical strength and clarity of constraint-based methods. We use a Bayesian score to obtain probability estimates on the input statements used in a constraint-based procedure. These are subsequently processed in decreasing order of reliability, letting more reliable decisions take precedence in case of conflicts, until a single output model is obtained. Tests show that a basic implementation of the resulting Bayesian Constraint-based Causal Discovery (BCCD) algorithm already outperforms established procedures such as FCI and Conservative PC. It indicates which causal decisions in the output have high reliability and which do not. The approach is easily adapted to other application areas such as complex independence tests.

Thresholding a measure in conditional independence (CI) tests using a fixed value enables learning and removing edges as part of learning a Bayesian network structure. However, the learned structure is sensitive to the threshold that is commonly selected: 1) arbitrarily; 2) irrespective of characteristics of the domain; and 3) fixed for all CI tests. We analyze the impact on mutual information – a CI measure – of factors, such as sample size, degree of variable dependence, and variables’ cardinalities. Following, we suggest to adaptively threshold individual tests based on the factors. We show that adaptive thresholds better distinguish between pairs of dependent variables and pairs of independent variables and enable learning structures more accurately and quickly than when using fixed thresholds.