We describe a method for identifying the speakers of quoted speech in natural-language textual stories. We have assembled a corpus of more than 3,000 quotations, whose speakers (if any) are manually identified, from a collection of 19th and 20th century literature by six authors. Using rule-based and statistical learning, our method identifies candidate characters, determines their genders, and attributes each quote to the most likely speaker. We divide the quotes into syntactic classes in order to leverage common discourse patterns, which enable rapid attribution for many quotes. We apply learning algorithms to the remainder and achieve an overall accuracy of 83%.

Due to the non-stationary environment, learning in multi-agent systems is a challenging problem. This paper first introduces a new gradient-based learning algorithm, augmenting the basic gradient ascent approach with policy prediction. We prove that this augmentation results in a stronger notion of convergence than the basic gradient ascent, that is, strategies converge to a Nash equilibrium within a restricted class of iterated games. Motivated by this augmentation, we then propose a new practical multi-agent reinforcement learning (MARL) algorithm exploiting approximate policy prediction. Empirical results show that it converges faster and in a wider variety of situations than state-of-the-art MARL algorithms.

Trial-based approaches offer an efficient way to solve single-agent MDPs and POMDPs. These approaches allow agents to focus their computations on regions of the environment they encounter during the trials, leading to significant computational savings. We present a novel trial-based dynamic programming (TBDP) algorithm for DEC-POMDPs that extends these benefits to multi-agent settings. The algorithm uses trial-based methods for both belief generation and policy evaluation. Policy improvement is implemented efficiently using linear programming and a sub-policy reuse technique that helps bound the amount of memory. The results show that TBDP can produce significant value improvements and is much faster than the best existing planning algorithms.

We present an on-line decentralised algorithm for coordinating mobile sensors for a broad class of information gathering tasks. These sensors can be deployed in unknown and possibly hostile environments, where uncertainty and dynamism are endemic. Such environments are common in the areas of disaster response and military surveillance. Our coordination approach itself is based on work by Stranders et al. (2009), that uses the max-sum algorithm to coordinate mobile sensors for monitoring spatial phenomena. In particular, we generalise and extend their approach to any domain where measurements can be valued. Also, we introduce a clustering approach that allows sensors to negotiate over paths to the most relevant locations, as opposed to a set of fixed directions, which results in a significantly improved performance. We demonstrate our algorithm by applying it to two challenging and distinct information gathering tasks. In the first–pursuit-evasion (PE)–sensors need to capture a target whose movement might be unknown. In the second–patrolling (P)–sensors need to minimise loss from intrusions that occur within their environment. In doing so, we obtain the first decentralised coordination algorithms for these domains. Finally, in each domain, we empirically evaluate our approach in a simulated environment, and show that it outperforms two state of the art greedy algorithms by 30% (PE) and 44% (P), and an existing approach based on the Travelling Salesman Problem by 52% (PE) and 30% (P).

Using a model of agent behavior based around envy-reducing strategies, we describe an iterated combinatorial auction in which the allocation and prices converge to a solution in the core of the agents' true valuations. In each round of the iterative auction mechanism, agents act on envy quotes produced by the mechanism: hints that suggest the prices of the bundles they are interested in. We describe optimal methods of generating envy quotes for various core-selecting mechanisms. Prior work on core-selecting combinatorial auctions has required agents to have perfect information about every agent's valuations to achieve a solution in the core. In contrast, here a core solution is reached even in the private information setting.

Multi-agent decision problems, in which independent agents have to agree on a joint plan of action or allocation of resources, are central to AI. In such situations, agents' individual preferences over available alternatives may vary, and they may try to reconcile these differences by voting. Based on the fact that agents may have incentives to vote strategically and misreport their real preferences, a number of recent papers have explored different possibilities for avoiding or eliminating such manipulations. In contrast to most prior work, this paper focuses on convergence of strategic behavior to a decision from which no voter will want to deviate. We consider scenarios where voters cannot coordinate their actions, but are allowed to change their vote after observing the current outcome. We focus on the Plurality voting rule, and study the conditions under which this iterative game is guaranteed to converge to a Nash equilibrium (i.e., to a decision that is stable against further unilateral manipulations). We show for the first time how convergence depends on the exact attributes of the game, such as the tie-breaking scheme, and on assumptions regarding agents' weights and strategies.

Spectral feature selection identifies relevant features by measuring their capability of preserving sample similarity. It provides a powerful framework for both supervised and unsupervised feature selection, and has been proven to be effective in many real-world applications. One common drawback associated with most existing spectral feature selection algorithms is that they evaluate features individually and cannot identify redundant features. Since redundant features can have significant adverse effect on learning performance, it is necessary to address this limitation for spectral feature selection. To this end, we propose a novel spectral feature selection algorithm to handle feature redundancy, adopting an embedded model. The algorithm is derived from a formulation based on a sparse multi-output regression with a L 2,1 -norm constraint. We conduct theoretical analysis on the properties of its optimal solutions, paving the way for designing an efficient path-following solver. Extensive experiments show that the proposed algorithm can do well in both selecting relevant features and removing redundancy.

We present an interactive learning method that enables a user to iteratively refine a regression model. The user examines the output of the model, visualized as the vertical axis of a 2D scatterplot, and provides corrections by repositioning individual data instances to the correct output level. Each repositioned data instance acts as a control point for altering the learned model, using the geometry underlying the data. We capture the underlying structure of the data as a manifold, on which we compute a set of basis functions as the foundation for learning. Our results show that manifold-based interactive learning improves performance monotonically with each correction, outperforming alternative approaches.

Interactive clustering refers to situations in which a human labeler is willing to assist a learning algorithm in automatically clustering items. We present a related but somewhat different task, assisted clustering, in which a user creates explicit groups of items from a large set and wants suggestions on what items to add to each group. While the traditional approach to interactive clustering has been to use metric learning to induce a distance metric, our situation seems equally amenable to classification. Using clusterings of documents from human subjects, we found that one or the other method proved to be superior for a given cluster, but not uniformly so. We thus developed a hybrid mechanism for combining the metric learner and the classifier. We present results from a large number of trials based on human clusterings, in which we show that our combination scheme matches and often exceeds the performance of a method which exclusively uses either type of learner.

Belief revision performs belief change on an agent's beliefs when new evidence (either of the form of a propositional formula or of the form of a total pre-order on a set of interpretations) is received. Jeffrey's rule is commonly used for revising probabilistic epistemic states when new information is probabilistically uncertain. In this paper, we propose a general epistemic revision framework where new evidence is of the form of a partial epistemic state. Our framework extends Jeffrey's rule with uncertain inputs and covers well-known existing frameworks such as ordinal conditional function (OCF) or possibility theory. We then define a set of postulates that such revision operators shall satisfy and establish representation theorems to characterize those postulates. We show that these postulates reveal common characteristics of various existing revision strategies and are satisfied by OCF conditionalization, Jeffrey's rule of conditioning and possibility conditionalization. Furthermore, when reducing to the belief revision situation, our postulates can induce most of Darwiche and Pearl's postulates.