Classification trees are widely used in the machine learning and data mining communities for modeling propositional data. Recent work has extended this basic paradigm to probability estimation trees. Traditional tree learning algorithms assume that instances in the training data are homogenous and independently distributed. Relational probability trees (RPTs) extend standard probability estimation trees to a relational setting in which data instances are heterogeneous and interdependent. Our algorithm for learning the structure and parameters of an RPT searches over a space of relational features that use aggregation functions (e.g. AVERAGE, MODE, COUNT) to dynamically propositionalize relational data and create binary splits within the RPT. Previous work has identified a number of statistical biases due to characteristics of relational data such as autocorrelation and degree disparity. The RPT algorithm uses a novel form of randomization test to adjust for these biases. On a variety of relational learning tasks, RPTs built using randomization tests are significantly smaller than other models and achieve equivalent, or better, performance. 1.

Propositionalization has already been shown to be a promising approach for robustly and e#ectively handling relational data sets for knowledge discovery. In this paper, we compare up-to-date methods for propositionalization from two main groups: logic-oriented and databaseoriented techniques. Experiments using several learning tasks --- both ILP benchmarks and tasks from recent international data mining competitions --- show that both groups have their specific advantages. While logic-oriented methods can handle complex background knowledge and provide expressive first-order models, database-oriented methods can be more e#cient especially on larger data sets. Obtained accuracies vary such that a combination of the features produced by both groups seems a further valuable venture.

To learn to behave in highly complex domains, agents must represent and learn compact models of the world dynamics. In this paper, we present an algorithm for learning probabilistic STRIPS-like planning operators from examples. We demonstrate the effective learning of rule-based operators for a wide range of traditional planning domains.

Feature construction through aggregation plays an essential role in modeling relational domains with one-to-many relationships between tables. One-to-many relationships lead to bags (multisets) of related entities, from which predictive information must be captured. This paper focuses on aggregation from categorical attributes that can take many values (e.g., object identifiers). We present a novel aggregation method as part of a relational learning system ACORA, that combines the use of vector distance and meta-data about the class-conditional distributions of attribute values. We provide a theoretical foundation for this approach deriving a “relational fixed-effect ” model within a Bayesian framework, and discuss the implications of identifier aggregation on the expressive power of the induced model. One advantage of using identifier attributes is the circumvention of limitations caused either by missing/unobserved object properties or by independence assumptions. Finally, we show empirically that the novel aggregators can generalize in the presence of identifier (and other high-dimensional) attributes, and also explore the limitations of the applicability of the methods. 1

Abstract. Conditional Random Fields (CRFs) provide a powerful instrument for labeling sequences. So far, however, CRFs have only been considered for labeling sequences over flat alphabets. In this paper, we describe TildeCRF, the first method for training CRFs on logical sequences, i.e., sequences over an alphabet of logical atoms. TildeCRF’s key idea is to use relational regression trees in Dietterich et al.’s gradient tree boosting approach. Thus, the CRF potential functions are represented as weighted sums of relational regression trees. Experiments show a significant improvement over established results achieved with hidden Markov models and Fisher kernels for logical sequences. 1

Stochastic processes that involve the creation of objects and relations over time are widespread, but relatively poorly studied. For example, accurate fault diagnosis in factory assembly processes requires inferring the probabilities of erroneous assembly operations, but doing this efficiently and accurately is difficult. Modeled as dynamic Bayesian networks, these processes have discrete variables with very large domains and extremely high dimensionality. In this paper, we introduce relational dynamic Bayesian networks (RDBNs), which are an extension of dynamic Bayesian networks (DBNs) to first-order logic. RDBNs are a generalization of dynamic probabilistic relational models (DPRMs), which we had proposed in our previous work to model dynamic uncertain domains. We first extend the Rao-Blackwellised particle filtering described in our earlier work to RDBNs. Next, we lift the assumptions associated with Rao-Blackwellization in RDBNs and propose two new forms of particle filtering. The first one uses abstraction hierarchies over the predicates to smooth the particle filter’s estimates. The second employs kernel density estimation with a kernel function specifically designed for relational domains. Experiments show these two methods greatly outperform standard particle filtering on the task of assembly plan execution monitoring. 1.

We consider the task of tagging Slovene words with morphosyntactic descriptions (MSDs). MSDs contain not only part-of-speech information but also attributes such as gender and case. In the case of Slovene there are 2,083 possible MSDs. P-Progol was used to learn morphosyntactic disambiguation rules from annotated data (consisting of 161,314 examples) produced by the MULTEXT-East project. P-Progol produced 1,148 rules taking 36 hours. Using simple grammatical background knowledge, e.g. looking for case disagreement, P-Progol induced 4,094 clauses in eight parallel runs. These rules have proved effective at detecting and explaining incorrect MSD annotations in an independent test set, but have not so far produced a tagger comparable to other existing taggers in terms of accuracy.

This paper deals with the issue of learning in multi-agent systems (MAS). Particularly, we are interested in BDI (Belief, Desire, Intention) agents. Despite the relevance of the BDI model of rational agency, little work has been done to deal with its two main limitations: i) The lack of learning competences; and ii) The lack of explicit multi-agent functionality. From the

The current evaluation functions for heuristic planning are expensive to compute. In numerous domains these functions give good guidance on the solution, so it worths the computation effort. On the contrary, where this is not true, heuristics planners compute loads of useless node evaluations that make them scale-up poorly. In this paper we present a novel approach for boosting the scalability of heuristic planners based on automatically learning domain-specific search control knowledge in the form of relational decision trees. Particularly, we define the learning of planning search control as a standard classification process. Then, we use an off-theshelf relational classifier to build domain-specific relational decision trees that capture the preferred action in the different planning contexts of a planning domain. These contexts are defined by the set of helpful actions extracted from the relaxed planning graph of a given state, the goals remaining to be achieved, and the static predicates of the planning task. Additionally, we show two methods for guiding the search of a heuristic planner with relational decision trees. The first one consists of using the resulting decision trees as an action policy. The second one consists of ordering the node evaluation of the Enforced Hill Climbing algorithm with the learned decision trees. Experiments over a variety of domains from the IPC test-benchmarks reveal that in both cases the use of the learned decision trees increase the number of problems solved together with a reduction of the time spent.

Abstract. Dependency networks approximate a joint probability distribution over multiple random variables as a product of conditional distributions. Relational Dependency Networks (RDNs) are graphical models that extend dependency networks to relational domains. This higher expressivity, however, comes at the expense of a more complex modelselection problem: an unbounded number of relational abstraction levels might need to be explored. Whereas current learning approaches for RDNs learn a single probability tree per random variable, we propose to turn the problem into a series of relational function-approximation problems using gradient-based boosting. In doing so, one can easily induce highly complex features over several iterations and in turn estimate quickly a very expressive model. Our experimental results in several different data sets show that this boosting method results in efficient learning of RDNs when compared to state-of-the-art statistical relational learning approaches. 1