This document is a thesis on the subject of single-agent on-line path planning in continuous,unpredictable and highly dynamic environments. The problem is finding and traversing a collision-free path for a holonomic robot, without kinodynamic restrictions, moving in an environment with several unpredictably moving obstacles or adversaries. The availability of perfect information of the environment at all times is assumed. Several static and dynamic variants of the Rapidly Exploring Random Trees (RRT) algorithm are explored, as well as an evolutionary algorithm for planning in dynamic environments called the Evolutionary Planner/Navigator. A combination of both kinds of algorithms is proposed to overcome shortcomings in both, and then a combination of a RRT variant for initial planning and informed local search for navigation, plus a simple greedy heuristic for optimization. We show that this combination of simple techniques provides better responses to highly dynamic environments than the RRT extensions.

Probabilistic sampling methods have become very popular to solve single-shot path planning problems. Rapidly-exploring Random Trees (RRTs) in particular have been shown to be efficient in solving high dimensional problems. Even though several RRT variants have been proposed for dynamic replanning, these methods only perform well in environments with infrequent changes. This paper addresses the dynamic path planning problem by combining simple techniques in a multi-stage probabilistic algorithm. This algorithm uses RRTs for initial planning and informed local search for navigation. We show that this combination of simple techniques provides better responses to highly dynamic environments than the RRT extensions.

Adaptation has long been considered as the Achilles' heel of case-based reasoning since it requires some domain-specific knowledge that is difficult to acquire. In this paper, two strategies are combined in order to reduce the knowledge engineering cost induced by the adaptation knowledge (AK) acquisition task: AK is learned from the case base by the means of knowledge discovery techniques, and the AK acquisition sessions are opportunistically triggered, i.e., at problem-solving time.

We propose new definitions of (causal) explanation, using structural equations to model counterfactuals. The definition is based on the notion of actual cause, as defined and motivated in a companion paper. Essentially, an explanation is a fact that is not known for certain but, if found to be true, would constitute an actual cause of the fact to be explained, regardless of the agent's initial uncertainty. We show that the definition handles well a number of problematic examples from the literature.

We propose a new definition of actual cause, using structural equations to model counterfactuals. We show that the definition yields a plausible and elegant account of causation that handles well examples which have caused problems for other definitions and resolves major difficulties in the traditional account.

Reinforcement learning is commonly used with function approximation. However, very few positive results are known about the convergence of function approximation based RL control algorithms. In this paper we show that TD(0) and Sarsa(0) with linear function approximation is convergent for a simple class of problems, where the system is linear and the costs are quadratic (the LQ control problem). Furthermore, we show that for systems with Gaussian noise and non-completely observable states (the LQG problem), the mentioned RL algorithms are still convergent, if they are combined with Kalman filtering.

This paper is about automatic acquisition of lexical information from corpora, especially subcategorization acquisition.

Directed acyclic graphs (DAGs) are commonly used to represent causal relationships among random variables in graphical models. Applications of these models arise in the study of physical, as well as biological systems, where directed edges between nodes represent the influence of components of the system on each other. The general problem of estimating DAGs from observed data is computationally NP-hard, Moreover two directed graphs may be observationally equivalent. When the nodes exhibit a natural ordering, the problem of estimating directed graphs reduces to the problem of estimating the structure of the network. In this paper, we propose a penalized likelihood approach that directly estimates the adjacency matrix of DAGs. Both lasso and adaptive lasso penalties are considered and an efficient algorithm is proposed for estimation of high dimensional DAGs. We study variable selection consistency of the two penalties when the number of variables grows to infinity with the sample size. We show that although lasso can only consistently estimate the true network under stringent assumptions, adaptive lasso achieves this task under mild regularity conditions. The performance of the proposed methods is compared to alternative methods in simulated, as well as real, data examples.

Recently there has been an increasing interest in methods that deal with multiple outputs. This has been motivated partly by frameworks like multitask learning, multisensor networks or structured output data. From a Gaussian processes perspective, the problem reduces to specifying an appropriate covariance function that, whilst being positive semi-definite, captures the dependencies between all the data points and across all the outputs. One approach to account for non-trivial correlations between outputs employs convolution processes. Under a latent function interpretation of the convolution transform we establish dependencies between output variables. The main drawbacks of this approach are the associated computational and storage demands. In this paper we address these issues. We present different sparse approximations for dependent output Gaussian processes constructed through the convolution formalism. We exploit the conditional independencies present naturally in the model. This leads to a form of the covariance similar in spirit to the so called PITC and FITC approximations for a single output. We show experimental results with synthetic and real data, in particular, we show results in pollution prediction, school exams score prediction and gene expression data.

In a recent work (arXiv:0910.2517), for nonlinear models with sparse underlying linear structures, we studied the error bounds of $\ell_0$-regularized estimation. In this note, we show that $\ell_1$-regularized estimation in some important cases can achieve the same order of error bounds as those in the aforementioned work.