We study the problem of learning probabilistic models of high-level strategic behavior in the real-time strategy (RTS) game StarCraft. The models are automatically learned from sets of game logs and aim to capture the common strategic states and decision points that arise in those games. Unlike most work on behavior/strategy learning and prediction in RTS games, our data-centric approach is not biased by or limited to any set of preconceived strategic concepts. Further, since our behavior model is based on the well-developed and generic paradigm of hidden Markov models, it supports a variety of uses for the design of AI players and human assistants. For example, the learned models can be used to make probabilistic predictions of a player's future actions based on observations, to simulate possible future trajectories of a player, or to identify uncharacteristic or novel strategies in a game database. In addition, the learned qualitative structure of the model can be analyzed by humans in order to categorize common strategic elements. We demonstrate our approach by learning models from 331 expert-level games and provide both a qualitative and quantitative assessment of the learned model's utility.

Interactive narrative environments offer significant potential for creating engaging narrative experiences that are tailored to individual users. Increasingly, applications in education, training, and entertainment are leveraging narrative to create rich interactive experiences in virtual storyworlds. A key challenge posed by these environments is devising accurate models of director agents’ strategies that determine the most appropriate director action to perform for crafting customized story experiences. A promising approach is developing an empirically informed model of director agents’ decision-making strategies. In this paper, we propose a framework for learning models of director agent decision-making strategies by observing human-human interactions in an interactive narrative-centered learning environment. The results are encouraging and suggest that creating empirically driven models of director agent decision-making is a promising approach to interactive narrative.

We consider the problem of learning the structure of Ising models (pairwise binary Markov random fields) from i.i.d. samples. While several methods have been proposed to accomplish this task, their relative merits and limitations remain somewhat obscure. By analyzing a number of concrete examples, we show that low-complexity algorithms often fail when the Markov random field develops long-range correlations. More precisely, this phenomenon appears to be related to the Ising model phase transition (although it does not coincide with it).

In this paper we propose a data intensive approach for inferring sentence-internal temporal relations. Temporal inference is relevant for practical NLP applications which either extract or synthesize temporal information (e.g., summarisation, question answering). Our method bypasses the need for manual coding by exploiting the presence of markers like after", which overtly signal a temporal relation. We first show that models trained on main and subordinate clauses connected with a temporal marker achieve good performance on a pseudo-disambiguation task simulating temporal inference (during testing the temporal marker is treated as unseen and the models must select the right marker from a set of possible candidates). Secondly, we assess whether the proposed approach holds promise for the semi-automatic creation of temporal annotations. Specifically, we use a model trained on noisy and approximate data (i.e., main and subordinate clauses) to predict intra-sentential relations present in TimeBank, a corpus annotated rich temporal information. Our experiments compare and contrast several probabilistic models differing in their feature space, linguistic assumptions and data requirements. We evaluate performance against gold standard corpora and also against human subjects.

This thesis derives, tests and applies two linear projection algorithms for machine learning under non-stationarity. The first finds a direction in a linear space upon which a data set is maximally non-stationary. The second aims to robustify two-way classification against non-stationarity. The algorithm is tested on a key application scenario, namely Brain Computer Interfacing.

The Partially Observable Markov Decision Process has long been recognized as a rich framework for real-world planning and control problems, especially in robotics. However exact solutions in this framework are typically computationally intractable for all but the smallest problems. A well-known technique for speeding up POMDP solving involves performing value backups at specific belief points, rather than over the entire belief simplex. The efficiency of this approach, however, depends greatly on the selection of points. This paper presents a set of novel techniques for selecting informative belief points which work well in practice. The point selection procedure is combined with point-based value backups to form an effective anytime POMDP algorithm called Point-Based Value Iteration (PBVI). The first aim of this paper is to introduce this algorithm and present a theoretical analysis justifying the choice of belief selection technique. The second aim of this paper is to provide a thorough empirical comparison between PBVI and other state-of-the-art POMDP methods, in particular the Perseus algorithm, in an effort to highlight their similarities and differences. Evaluation is performed using both standard POMDP domains and realistic robotic tasks.

Standard value function approaches to finding policies for Partially Observable Markov Decision Processes (POMDPs) are generally considered to be intractable for large models. The intractability of these algorithms is to a large extent a consequence of computing an exact, optimal policy over the entire belief space. However, in real-world POMDP problems, computing the optimal policy for the full belief space is often unnecessary for good control even for problems with complicated policy classes. The beliefs experienced by the controller often lie near a structured, low-dimensional subspace embedded in the high-dimensional belief space. Finding a good approximation to the optimal value function for only this subspace can be much easier than computing the full value function. We introduce a new method for solving large-scale POMDPs by reducing the dimensionality of the belief space. We use Exponential family Principal Components Analysis (Collins, Dasgupta and Schapire, 2002) to represent sparse, high-dimensional belief spaces using small sets of learned features of the belief state. We then plan only in terms of the low-dimensional belief features. By planning in this low-dimensional space, we can find policies for POMDP models that are orders of magnitude larger than models that can be handled by conventional techniques. We demonstrate the use of this algorithm on a synthetic problem and on mobile robot navigation tasks.

Modelling the real world complexity of music is a challenge for machine learning. We address the task of modeling melodic sequences from the same music genre. We perform a comparative analysis of two probabilistic models; a Dirichlet Variable Length Markov Model (Dirichlet-VMM) and a Time Convolutional Restricted Boltzmann Machine (TC-RBM). We show that the TC-RBM learns descriptive music features, such as underlying chords and typical melody transitions and dynamics. We assess the models for future prediction and compare their performance to a VMM, which is the current state of the art in melody generation. We show that both models perform significantly better than the VMM, with the Dirichlet-VMM marginally outperforming the TC-RBM. Finally, we evaluate the short order statistics of the models, using the Kullback-Leibler divergence between test sequences and model samples, and show that our proposed methods match the statistics of the music genre significantly better than the VMM.

Kernel methods have long provided powerful tools for generalizing linear statistical approaches to nonlinear settings, through an embedding of the sample to a high dimensional feature space, namely a reproducing kernel Hilbert space (RKHS) [18, 28]. Examples include support vector machines, kernel PCA, and kernel CCA, among others. In these cases, data are mapped via a canonical feature map to a reproducing kernel Hilbert space (of high or even infinite dimension), in which the linear operations that define the algorithms are implemented. The inner product between feature mappings need never be computed explicitly, but is given by a positive definite kernel function unique to the RKHS: this permits efficient computation without the need to deal explicitly with the feature representation. The mappings of individual points to a feature space may be generalized to mappings of probability measures[e.g. 3, Chapter 4]. We call such mappings the kernel means of the underlying random variables.

Open distributed multi-agent systems are gaining interest in the academic community and in industry. In such open settings, agents are often coordinated using standardized agent conversation protocols. The representation of such protocols (for analysis, validation, monitoring, etc) is an important aspect of multi-agent applications. Recently, Petri nets have been shown to be an interesting approach to such representation, and radically different approaches using Petri nets have been proposed. However, their relative strengths and weaknesses have not been examined. Moreover, their scalability and suitability for different tasks have not been addressed. This paper addresses both these challenges. First, we analyze existing Petri net representations in terms of their scalability and appropriateness for overhearing, an important task in monitoring open multi-agent systems. Then, building on the insights gained, we introduce a novel representation using Colored Petri nets that explicitly represent legal joint conversation states and messages. This representation approach offers significant improvements in scalability and is particularly suitable for overhearing. Furthermore, we show that this new representation offers a comprehensive coverage of all conversation features of FIPA conversation standards. We also present a procedure for transforming AUML conversation protocol diagrams (a standard human-readable representation), to our Colored Petri net representation.