Context-specific Independence Research Articles

A new class of generative classifiers based on staged tree models

Generative models for classification use the joint probability distribution of the class variable and the features to construct a decision rule. Among generative models, Bayesian networks and naive Bayes classifiers are the most commonly used and provide a clear graphical representation of the relationship among all variables. However, these have the disadvantage of highly restricting the type of relationships that could exist, by not allowing for context-specific independence. Here we introduce a new class of generative classifiers, called staged tree classifiers, which formally account for context-specific independence. They are constructed by a partitioning of the vertices of an event tree from which conditional independence can be formally read. The naive staged tree classifier is also defined, which extends the classic naive Bayes classifier whilst retaining the same complexity. An extensive simulation study shows that the classification accuracy of staged tree classifiers is competitive with that of state-of-the-art classifiers and an example showcases their use in practice.

Value‐based potentials: Exploiting quantitative information regularity patterns in probabilistic graphical models

When dealing with complex models (i.e., models with many variables, a high degree of dependency between variables, or many states per variable), the efficient representation of quantitative information in probabilistic graphical models (PGMs) is a challenging task. To address this problem, this study introduces several new structures, aptly named value-based potentials (VBPs), which are based exclusively on the values. VBPs leverage repeated values to reduce memory requirements. In the present paper, they are compared with some common structures, like standard tables or unidimensional arrays, and probability trees (PT). Like VBPs, PTs are designed to reduce the memory space, but this is achieved only if value repetitions correspond to context-specific independence patterns (i.e., repeated values are related to consecutive indices or configurations). VBPs are devised to overcome this limitation. The goal of this study is to analyze the properties of VBPs. We provide a theoretical analysis of VBPs and use them to encode the quantitative information of a set of well-known Bayesian networks, measuring the access time to their content and the computational time required to perform some inference tasks.

Sum–product graphical models

This paper introduces a probabilistic architecture called sum–product graphical model (SPGM). SPGMs represent a class of probability distributions that combines, for the first time, the semantics of probabilistic graphical models (GMs) with the evaluation efficiency of sum–product networks (SPNs): Like SPNs, SPGMs always enable tractable inference using a class of models that incorporate context specific independence. Like GMs, SPGMs provide a high-level model interpretation in terms of conditional independence assumptions and corresponding factorizations. Thus, this approach provides new connections between the fields of SPNs and GMs, and enables a high-level interpretation of the family of distributions encoded by SPNs. We provide two applications of SPGMs in density estimation with empirical results close to or surpassing state-of-the-art models. The theoretical and practical results demonstrate that jointly exploiting properties of SPNs and GMs is an interesting direction of future research.

A logical approach to context-specific independence

Directed acyclic graphs (DAGs) constitute a qualitative representation for conditional independence (CI) properties of a probability distribution. It is known that every CI statement implied by the topology of a DAG is witnessed over it under a graph-theoretic criterion of d-separation. Alternatively, all such implied CI statements are derivable from the local independencies encoded by a DAG using the so-called semi-graphoid axioms. We consider Labeled Directed Acyclic Graphs (LDAGs) modeling graphically scenarios exhibiting context-specific independence (CSI). Such CSI statements are modeled by labeled edges, where labels encode contexts in which the edge vanishes. We study the problem of identifying all independence statements implied by the structure and the labels of an LDAG. We show that this problem is coNP-hard for LDAGs and formulate a sound extension of the semi-graphoid axioms for the derivation of such implied independencies. Finally we connect our study to certain qualitative versions of independence ubiquitous in database theory and teams semantics.

Exact Bayesian learning of Partition Directed Acyclic Graphs

It is known that directed acyclic graphs (DAGs) may hide several local features of the joint probability distribution that can be essential for some applications. To remedy this, more expressive model classes have been introduced. In addition to the restrictions implied by conditional independence, these model classes typically include some form of local structure that implies equality constraints on the node-wise conditional distribution. In particular, the concept of context-specific independence (CSI) was introduced to increase the flexibility of traditional Bayesian networks. Furthermore, in the most expressive class of generalized Bayesian networks, decision graphs were used to model arbitrary parameter restrictions. Here we formulate an alternative representation of such models called a partition DAG (PDAG), which defines the parameter restrictions using a partition-based representation of the parent outcome spaces. We establish a criterion that can identify whether an arbitrary PDAG has a CSI-consistent representation using an efficient basic graph theoretic algorithm. Based on a recursive inference algorithm for partition posteriors, an exact Bayesian learning method is introduced. We demonstrate on real data that exact learning of PDAGs can identify important relationships between variables that have not been discovered by previous graphical model learning methods.

Decision-Theoretic Planning Under Anonymity in Agent Populations

We study the problem of self-interested planning under uncertainty in settings shared with more than a thousand other agents, each of which plans at its own individual level. We refer to such large numbers of agents as an agent population. The decision-theoretic formalism of interactive partially observable Markov decision process (I-POMDP) is used to model the agent's self-interested planning. The first contribution of this article is a method for drastically scaling the finitely-nested I-POMDP to certain agent populations for the first time. Our method exploits two types of structure that is often exhibited by agent populations -- anonymity and context-specific independence. We present a variant called the many-agent I-POMDP that models both these types of structure to plan efficiently under uncertainty in multiagent settings. In particular, the complexity of the belief update and solution in the many-agent I-POMDP is polynomial in the number of agents compared with the exponential growth that challenges the original framework. While exploiting structure helps mitigate the curse of many agents, the well-known curse of history that afflicts I-POMDPs continues to challenge scalability in terms of the planning horizon. The second contribution of this article is an application of the branch-and-bound scheme to reduce the exponential growth of the search tree for look ahead. For this, we introduce new fast-computing upper and lower bounds for the exact value function of the many-agent I-POMDP. This speeds up the look-ahead computations without trading off optimality, and reduces both memory and run time complexity. The third contribution is a comprehensive empirical evaluation of the methods on three new problems domains -- policing large protests, controlling traffic congestion at a busy intersection, and improving the AI for the popular Clash of Clans multiplayer game. We demonstrate the feasibility of exact self-interested planning in these large problems, and that our methods for speeding up the planning are effective. Altogether, these contributions represent a principled and significant advance toward moving self-interested planning under uncertainty to real-world applications.

Latent Dependency Forest Models

Probabilistic modeling is one of the foundations of modern machine learning and artificial intelligence. In this paper, we propose a novel type of probabilistic models named latent dependency forest models (LDFMs). A LDFM models the dependencies between random variables with a forest structure that can change dynamically based on the variable values. It is therefore capable of modeling context-specific independence. We parameterize a LDFM using a first-order non-projective dependency grammar. Learning LDFMs from data can be formulated purely as a parameter learning problem, and hence the difficult problem of model structure learning is circumvented. Our experimental results show that LDFMs are competitive with existing probabilistic models.

Stratified Gaussian graphical models

ABSTRACTGaussian graphical models represent the backbone of the statistical toolbox for analyzing continuous multivariate systems. However, due to the intrinsic properties of the multivariate normal distribution, use of this model family may hide certain forms of context-specific independence that are natural to consider from an applied perspective. Such independencies have been earlier introduced to generalize discrete graphical models and Bayesian networks into more flexible model families. Here, we adapt the idea of context-specific independence to Gaussian graphical models by introducing a stratification of the Euclidean space such that a conditional independence may hold in certain segments but be absent elsewhere. It is shown that the stratified models define a curved exponential family, which retains considerable tractability for parameter estimation and model selection.

Exploiting local and repeated structure in Dynamic Bayesian Networks

We introduce the structural interface algorithm for exact probabilistic inference in Dynamic Bayesian Networks. It unifies state-of-the-art techniques for inference in static and dynamic networks, by combining principles of knowledge compilation with the interface algorithm. The resulting algorithm not only exploits the repeated structure in the network, but also the local structure, including determinism, parameter equality and context-specific independence. Empirically, we show that the structural interface algorithm speeds up inference in the presence of local structure, and scales to larger and more complex networks.

The role of local partial independence in learning of Bayesian networks

Bayesian networks are one of the most widely used tools for modeling multivariate systems. It has been demonstrated that more expressive models, which can capture additional structure in each conditional probability table (CPT), may enjoy improved predictive performance over traditional Bayesian networks despite having fewer parameters. Here we investigate this phenomenon for models of various degree of expressiveness on both extensive synthetic and real data. To characterize the regularities within CPTs in terms of independence relations, we introduce the notion of partial conditional independence (PCI) as a generalization of the well-known concept of context-specific independence (CSI). To model the structure of the CPTs, we use different graph-based representations which are convenient from a learning perspective. In addition to the previously studied decision trees and graphs, we introduce the concept of PCI-trees as a natural extension of the CSI-based trees. To identify plausible models we use the Bayesian score in combination with a greedy search algorithm. A comparison against ordinary Bayesian networks shows that models with local structures in general enjoy parametric sparsity and improved out-of-sample predictive performance, however, often it is necessary to regulate the model fit with an appropriate model structure prior to avoid overfitting in the learning process. The tree structures, in particular, lead to high quality models and suggest considerable potential for further exploration.

Context-specific independence in graphical log-linear models

Log-linear models are the popular workhorses of analyzing contingency tables. A log-linear parameterization of an interaction model can be more expressive than a direct parameterization based on probabilities, leading to a powerful way of defining restrictions derived from marginal, conditional and context-specific independence. However, parameter estimation is often simpler under a direct parameterization, provided that the model enjoys certain decomposability properties. Here we introduce a cyclical projection algorithm for obtaining maximum likelihood estimates of log-linear parameters under an arbitrary context-specific graphical log-linear model, which needs not satisfy criteria of decomposability. We illustrate that lifting the restriction of decomposability makes the models more expressive, such that additional context-specific independencies embedded in real data can be identified. It is also shown how a context-specific graphical model can correspond to a non-hierarchical log-linear parameterization with a concise interpretation. This observation can pave way to further development of non-hierarchical log-linear models, which have been largely neglected due to their believed lack of interpretability.

Individual Planning in Agent Populations: Exploiting Anonymity and Frame-Action Hypergraphs

Interactive partially observable Markov decision processes (I-POMDP) provide a formal framework for planning for a self-interested agent in multiagent settings. An agent operating in a multiagent environment must deliberate about the actions that other agents may take and the effect these actions have on the environment and the rewards it receives. Traditional I-POMDPs model this dependence on the actions of other agents using joint action and model spaces. Therefore, the solution complexity grows exponentially with the number of agents thereby complicating scalability. In this paper, we model and extend anonymity and context-specific independence — problem structures often present in agent populations — for computational gain. We empirically demonstrate the efficiency from exploiting these problem structures by solving a new multiagent problem involving more than 1,000 agents.

Labeled directed acyclic graphs: a generalization of context-specific independence in directed graphical models

We introduce a novel class of labeled directed acyclic graph (LDAG) models for finite sets of discrete variables. LDAGs generalize earlier proposals for allowing local structures in the conditional probability distribution of a node, such that unrestricted label sets determine which edges can be deleted from the underlying directed acyclic graph (DAG) for a given context. Several properties of these models are derived, including a generalization of the concept of Markov equivalence classes. Efficient Bayesian learning of LDAGs is enabled by introducing an LDAG-based factorization of the Dirichlet prior for the model parameters, such that the marginal likelihood can be calculated analytically. In addition, we develop a novel prior distribution for the model structures that can appropriately penalize a model for its labeling complexity. A non-reversible Markov chain Monte Carlo algorithm combined with a greedy hill climbing approach is used for illustrating the useful properties of LDAG models for both real and synthetic data sets.

Binary Cumulant Varieties

Algebraic statistics for binary random variables is concerned with highly structured algebraic varieties in the space of 2×2×···×2-tensors. We demonstrate the advantages of representing such varieties in the coordinate system of binary cumulants. Our primary focus lies on hidden subset models. Parametrizations and implicit equations in cumulants are derived for hyperdeterminants, for secant and tangential varieties of Segre varieties, and for certain context-specific independence models. Extending work of Rota and collaborators, we explore the polynomial inequalities satisfied by cumulants.

Modelling and inference with Conditional Gaussian Probabilistic Decision Graphs

Probabilistic Decision Graphs (PDGs) are probabilistic graphical models that represent a factorisation of a discrete joint probability distribution using a “decision graph”-like structure over local marginal parameters. The structure of a PDG enables the model to capture some context specific independence relations that are not representable in the structure of more commonly used graphical models such as Bayesian networks and Markov networks. This sometimes makes operations in PDGs more efficient than in alternative models. PDGs have previously been defined only in the discrete case, assuming a multinomial joint distribution over the variables in the model. We extend PDGs to incorporate continuous variables, by assuming a Conditional Gaussian (CG) joint distribution. We also show how inference can be carried out in an efficient way.

A Comprehensive Model of Audiovisual Perception: Both Percept and Temporal Dynamics

The sparse information captured by the sensory systems is used by the brain to apprehend the environment, for example, to spatially locate the source of audiovisual stimuli. This is an ill-posed inverse problem whose inherent uncertainty can be solved by jointly processing the information, as well as introducing constraints during this process, on the way this multisensory information is handled. This process and its result - the percept - depend on the contextual conditions perception takes place in. To date, perception has been investigated and modeled on the basis of either one of two of its dimensions: the percept or the temporal dynamics of the process. Here, we extend our previously proposed audiovisual perception model to predict both these dimensions to capture the phenomenon as a whole. Starting from a behavioral analysis, we use a data-driven approach to elicit a Bayesian network which infers the different percepts and dynamics of the process. Context-specific independence analyses enable us to use the model's structure to directly explore how different contexts affect the way subjects handle the same available information. Hence, we establish that, while the percepts yielded by a unisensory stimulus or by the non-fusion of multisensory stimuli may be similar, they result from different processes, as shown by their differing temporal dynamics. Moreover, our model predicts the impact of bottom-up (stimulus driven) factors as well as of top-down factors (induced by instruction manipulation) on both the perception process and the percept itself.

Action-Graph Games

Representing and reasoning with games becomes difficult once they involve large numbers of actions and players, because the space requirement for utility functions can grow unmanageably. Action-Graph Games (AGGs) are a fully-expressive game representation that can compactly express utility functions with structure such as context-specific independence, anonymity, and additivity. We show that AGGs can be used to compactly represent all games that are compact when represented as graphical games, symmetric games, anonymous games, congestion games, and polymatrix games, as well as games that require exponential space under all of these existing representations. We give a polynomial-time algorithm for computing a player's expected utility under an arbitrary mixed-strategy profile, and show how to use this algorithm to achieve exponential speedups of existing methods for computing sample Nash equilibria. We present results of experiments showing that using AGGs leads to a dramatic increase in the size of games accessible to computational analysis. 2 2 We gratefully acknowledge Moshe Tennenholtz for his co-authorship of a paper on Local Effect Games ( Leyton-Brown and Tennenholtz, 2003), an action-centric graphical model for games that inspired our work on AGGs.

Set-theoretic defining equations of the tangential variety of the Segre variety

We prove a set-theoretic version of the Landsberg–Weyman Conjecture on the defining equations of the tangential variety of a Segre product of projective spaces. We introduce and study the concept of exclusive rank. For the proof of this conjecture, we use a connection to the author’s previous work and re-express the tangential variety as the variety of principal minors of symmetric matrices that have exclusive rank no more than 1. We discuss applications to semiseparable matrices, tensor rank versus border rank, context-specific independence models and factor analysis models.

PyMix - The Python mixture package - a tool for clustering of heterogeneous biological data

BackgroundCluster analysis is an important technique for the exploratory analysis of biological data. Such data is often high-dimensional, inherently noisy and contains outliers. This makes clustering challenging. Mixtures are versatile and powerful statistical models which perform robustly for clustering in the presence of noise and have been successfully applied in a wide range of applications.ResultsPyMix - the Python mixture package implements algorithms and data structures for clustering with basic and advanced mixture models. The advanced models include context-specific independence mixtures, mixtures of dependence trees and semi-supervised learning. PyMix is licenced under the GNU General Public licence (GPL). PyMix has been successfully used for the analysis of biological sequence, complex disease and gene expression data.ConclusionsPyMix is a useful tool for cluster analysis of biological data. Due to the general nature of the framework, PyMix can be applied to a wide range of applications and data sets.

Partially-supervised protein subclass discovery with simultaneous annotation of functional residues

BackgroundThe study of functional subfamilies of protein domain families and the identification of the residues which determine substrate specificity is an important question in the analysis of protein domains. One way to address this question is the use of clustering methods for protein sequence data and approaches to predict functional residues based on such clusterings. The locations of putative functional residues in known protein structures provide insights into how different substrate specificities are reflected on the protein structure level.ResultsWe have developed an extension of the context-specific independence mixture model clustering framework which allows for the integration of experimental data. As these are usually known only for a few proteins, our algorithm implements a partially-supervised learning approach. We discover domain subfamilies and predict functional residues for four protein domain families: phosphatases, pyridoxal dependent decarboxylases, WW and SH3 domains to demonstrate the usefulness of our approach.ConclusionThe partially-supervised clustering revealed biologically meaningful subfamilies even for highly heterogeneous domains and the predicted functional residues provide insights into the basis of the different substrate specificities.