Markov logic networks (MLNs) combine the power of first-order logic and probabilistic graphical models and as a result are ideally suited for solving large, complex problems in application domains that have both rich relational structure and large amount of uncertainty. However, inference in these rich, relational representations is quite challenging. The aim of this thesis is to advance the state-of-the-art in MLN inference, enabling it to solve much harder and more complex tasks than is possible today. To this end, I will develop techniques that exploit logical structures and symmetries that are either explicitly or implicitly encoded in the MLN representation and demonstrate their usefulness by using them to solve hard real-world problems in the field of natural language understanding.

My dissertation research focuses on the application of hierarchical learning and heuristics based on social signals to solve challenges inherent to enabling human-robot collaboration. I approach this problem through advancing the state of the art in building hierarchical task representations, multi-agent task-level planning, and learning assistive behaviors from demonstration.

We present a probabilistic extension of logic programs under the stable model semantics, inspired by the concept of Markov Logic Networks. The proposed language takes advantage of both formalisms in a single framework, allowing us to represent commonsense reasoning problems that require both logical and probabilistic reasoning in an intuitive and elaboration tolerant way.

We provide a novel, flexible, iterative refinement algorithm to automatically construct an approximate statespace representation for Markov Decision Processes (MDPs). Our approach leverages bisimulation metrics, which have been used in prior work to generate features to represent the state space of MDPs.We address a drawback of this approach, which is the expensive computation of the bisimulation metrics. We propose an algorithm to generate an iteratively improving sequence of state space partitions. Partial metric computations guide the representation search and provide much lower space and computational complexity, while maintaining strong convergence properties. We provide theoretical results guaranteeing convergence as well as experimental illustrations of the accuracy and savings (in time and memory usage) of the new algorithm, compared to traditional bisimulation metric computation.

The main computational bottleneck in various sampling based and local-search based inference algorithms for Markov logic networks (e.g., Gibbs sampling, MC-SAT, MaxWalksat, etc.) is computing the number of groundings of a first-order formula that are true given a truth assignment to all of its ground atoms. We reduce this problem to the problem of counting the number of solutions of a constraint satisfaction problem (CSP) and show that during their execution, both sampling based and local-search based algorithms repeatedly solve dynamic versions of this counting problem. Deriving from the vast amount of literature on CSPs and graphical models, we propose an exact junction-tree based algorithm for computing the number of solutions of the dynamic CSP, analyze its properties, and show how it can be used to improve the computational complexity of Gibbs sampling and MaxWalksat. Empirical tests on a variety of benchmarks clearly show that our new approach is several orders of magnitude more scalable than existing approaches.

Lifted probabilistic inference algorithms have been successfully applied to a large number of symmetric graphical models. Unfortunately, the majority of real-world graphical models is asymmetric. This is even the case for relational representations when evidence is given. Therefore, more recent work in the community moved to making the models symmetric and then applying existing lifted inference algorithms. However, this approach has two shortcomings. First, all existing over-symmetric approximations require a relational representation such as Markov logic networks. Second, the induced symmetries often change the distribution significantly, making the computed probabilities highly biased. We present a framework for probabilistic sampling-based inference that only uses the induced approximate symmetries to propose steps in a Metropolis-Hastings style Markov chain. The framework, therefore, leads to improved probability estimates while remaining unbiased. Experiments demonstrate that the approach outperforms existing MCMC algorithms.

We provide a novel, flexible, iterative refinement algorithm to automatically construct an approximate statespace representation for Markov Decision Processes (MDPs). Our approach leverages bisimulation metrics, which have been used in prior work to generate features to represent the state space of MDPs. We address a drawback of this approach, which is the expensive computation of the bisimulation metrics. We propose an algorithm to generate an iteratively improving sequence of state space partitions. Partial metric computations guide the representation search and provide much lower space and computational complexity, while maintaining strong convergence properties. We provide theoretical results guaranteeing convergence as well as experimental illustrations of the accuracy and savings (in time and memory usage) of the new algorithm, compared to traditional bisimulation metric computation.

Two other algorithms require the knowledge Markov Decision Processes (MDPs) offer a general framework of the optimal policy and its expected reward. We show to describe probabilistic planning problems of varying that the expected reward of the optimal policy is a lower complexity. The development of algorithms that act successfully bound for the expected performance of both strategies. in MDPs is important to many AI applications. Our final algorithm switches between the application of Since it is often impossible or intractable to evaluate MDP the optimal policy and the policy with the highest possible algorithms based on a theoretical analysis alone, the International outcome, which can be computed without notable overhead Probabilistic Planning Competition (IPPC) was introduced in the Trial-based Heuristic Tree Search (THTS) framework to allow a comparison based on experimental evaluation. (Keller and Helmert 2013). We show theoretically and empirically The idea is to approximate the quality of an MDP that all algorithms outperform the naïve base approach solver by performing a sequence of runs on a problem instance, that ignores the potential of optimizing evaluation and by using the average of the obtained results as runs in hindsight, and that it pays off to take suboptimal base an approximation of the expected reward.

How should we gather information to make effective decisions? A classical answer to this fundamental problem is given by the decision-theoretic value of information. Unfortunately, optimizing this objective is intractable, and myopic (greedy) approximations are known to perform poorly. In this paper, we introduce DiRECt, an efficient yet near-optimal algorithm for nonmyopically optimizing value of information. Crucially, DiRECt uses a novel surrogate objective that is: (1) aligned with the value of information problem (2) efficient to evaluate and (3) adaptive submodular. This latter property enables us to utilize an efficient greedy optimization while providing strong approximation guarantees. We demonstrate the utility of our approach on four diverse case-studies: touch-based robotic localization, comparison-based preference learning, wild-life conservation management, and preference elicitation in behavioral economics. In the first application, we demonstrate DiRECt in closed-loop on an actual robotic platform.

We consider partially observable Markov decision processes (POMDPs) with a set of target states and every transition is associated with an integer cost. The optimization objective we study asks to minimize the expected total cost till the target set is reached, while ensuring that the target set is reached almost-surely (with probability 1). We show that for integer costs approximating the optimal cost is undecidable. For positive costs, our results are as follows: (i) we establish matching lower and upper bounds for the optimal cost and the bound is double exponential; (ii) we show that the problem of approximating the optimal cost is decidable and present approximation algorithms developing on the existing algorithms for POMDPs with finite-horizon objectives. While the worst-case running time of our algorithm is double exponential, we present efficient stopping criteria for the algorithm and show experimentally that it performs well in many examples of interest.