Cross-document coreference, the problem of resolving entity mentions across multi-document collections, is crucial to automated knowledge base construction and data mining tasks. However, the scarcity of large labeled data sets has hindered supervised machine learning research for this task. In this paper we develop and demonstrate an approach based on ``distantly-labeling'' a data set from which we can train a discriminative cross-document coreference model. In particular we build a dataset of more than a million people mentions extracted from 3.5 years of New York Times articles, leverage Wikipedia for distant labeling with a generative model (and measure the reliability of such labeling); then we train and evaluate a conditional random field coreference model that has factors on cross-document entities as well as mention-pairs. This coreference model obtains high accuracy in resolving mentions and entities that are not present in the training data, indicating applicability to non-Wikipedia data. Given the large amount of data, our work is also an exercise demonstrating the scalability of our approach.

Approximate dynamic programming has been used successfully in a large variety of domains, but it relies on a small set of provided approximation features to calculate solutions reliably. Large and rich sets of features can cause existing algorithms to overfit because of a limited number of samples. We address this shortcoming using $L_1$ regularization in approximate linear programming. Because the proposed method can automatically select the appropriate richness of features, its performance does not degrade with an increasing number of features. These results rely on new and stronger sampling bounds for regularized approximate linear programs. We also propose a computationally efficient homotopy method. The empirical evaluation of the approach shows that the proposed method performs well on simple MDPs and standard benchmark problems.

Aiming at accurate recognition of relations, we introduce local alignment kernels and explore various possibilities of using them for this task. We give a definition of a local alignment (LA) kernel based on the Smith-Waterman score as a sequence similarity measure and proceed with a range of possibilities for computing similarity between elements of sequences. We show how distributional similarity measures obtained from unlabeled data can be incorporated into the learning task as semantic knowledge. Our experiments suggest that the LA kernel yields promising results on various biomedical corpora outperforming two baselines by a large margin. Additional series of experiments have been conducted on the data sets of seven general relation types, where the performance of the LA kernel is comparable to the current state-of-the-art results.

We discuss an interaction-based approach to study the coevolution between socio-technical networks, individual behaviors, and contagion processes on these networks. We use epidemics in human population as an example of this phenomenon. The methods consist of developing synthetic yet realistic national-scale networks using a first principles approach. Unlike simple random graph techniques, these methods combine real world data sources with behavioral and social theories to synthesize detailed social contact (proximity) networks. Individual-based models of within-host disease progression and inter-host transmission are then used to model the contagion process. Finally, models of individual behaviors are composed with disease progression models to develop a realistic representation of the complex system in which individual behaviors and the social network adapt to the contagion. These methods are embodied within Simdemics – a general purpose modeling environment to support pandemic planning and response. Simdemics is designed specifically to be scalable to networks with 300 million agents – the underlying algorithms and methods in Simdemics are all high-performance computing oriented methods. New advances in network science, machine learning, high performance computing, data mining and behavioral modeling were necessary to develop Simdemics. Simdemics is combined with two other environments, Simfrastructure and Didactic, to form an integrated cyberenvironment. The integrated cyber-environment provides the end-user flexible and seamless Internet based access to Simdemics. Service-oriented architectures play a critical role in delivering the desired services to the end user. Simdemics, in conjunction with the integrated cyber-environment, has been used in over a dozen user defined case studies. These case studies were done to support specific policy questions that arose in the context of planning the response to pandemics (e.g., H1N1, H5N1) and human initiated bio-terrorism events. These studies played a crucial role in the continual development and improvement of the cyber-environment.

We develop a novel advanced Particle Markov chain Monte Carlo algorithm that is capable of sampling from the posterior distribution of non-linear state space models for both the unobserved latent states and the unknown model parameters. We apply this novel methodology to five population growth models, including models with strong and weak Allee effects, and test if it can efficiently sample from the complex likelihood surface that is often associated with these models. Utilising real and also synthetically generated data sets we examine the extent to which observation noise and process error may frustrate efforts to choose between these models. Our novel algorithm involves an Adaptive Metropolis proposal combined with an SIR Particle MCMC algorithm (AdPMCMC). We show that the AdPMCMC algorithm samples complex, high-dimensional spaces efficiently, and is therefore superior to standard Gibbs or Metropolis Hastings algorithms that are known to converge very slowly when applied to the non-linear state space ecological models considered in this paper. Additionally, we show how the AdPMCMC algorithm can be used to recursively estimate the Bayesian Cram\'er-Rao Lower Bound of Tichavsk\'y (1998). We derive expressions for these Cram\'er-Rao Bounds and estimate them for the models considered. Our results demonstrate a number of important features of common population growth models, most notably their multi-modal posterior surfaces and dependence between the static and dynamic parameters. We conclude by sampling from the posterior distribution of each of the models, and use Bayes factors to highlight how observation noise significantly diminishes our ability to select among some of the models, particularly those that are designed to reproduce an Allee effect.

Probabilistic databases play a crucial role in the management and understanding of uncertain data. However, incorporating probabilities into the semantics of incomplete databases has posed many challenges, forcing systems to sacrifice modeling power, scalability, or restrict the class of relational algebra formula under which they are closed. We propose an alternative approach where the underlying relational database always represents a single world, and an external factor graph encodes a distribution over possible worlds; Markov chain Monte Carlo (MCMC) inference is then used to recover this uncertainty to a desired level of fidelity. Our approach allows the efficient evaluation of arbitrary queries over probabilistic databases with arbitrary dependencies expressed by graphical models with structure that changes during inference. MCMC sampling provides efficiency by hypothesizing {\em modifications} to possible worlds rather than generating entire worlds from scratch. Queries are then run over the portions of the world that change, avoiding the onerous cost of running full queries over each sampled world. A significant innovation of this work is the connection between MCMC sampling and materialized view maintenance techniques: we find empirically that using view maintenance techniques is several orders of magnitude faster than naively querying each sampled world. We also demonstrate our system's ability to answer relational queries with aggregation, and demonstrate additional scalability through the use of parallelization.

Decentralized POMDPs are powerful theoretical models for coordinating agents’ decisions in uncertain environments, but the generally-intractable complexity of optimal joint policy construction presents a significant obstacle in applying Dec-POMDPs to problems where many agents face many policy choices. Here, we argue that when most agent choices are independent of other agents’ choices, much of this complexity can be avoided: instead of coordinating full policies, agents need only coordinate policy abstractions that explicitly convey the essential interaction influences. To this end, we develop a novel framework for influence-based policy abstraction for weakly-coupled transition-dependent Dec-POMDP problems that subsumes several existing approaches. In addition to formally characterizing the space of transition-dependent influences, we provide a method for computing optimal and approximately-optimal joint policies. We present an initial empirical analysis, over problems with commonly-studied flavors of transition-dependent influences, that demonstrates the potential computational benefits of influence-based abstraction over state-of-the-art optimal policy search methods.

Computing a good policy in stochastic uncertain environments with unknown dynamics and reward model parameters is a challenging task. In a number of domains, ranging from space robotics to epilepsy management, it may be possible to have an initial training period when suboptimal performance is permitted. For such problems it is important to be able to identify when this training period is complete, and the computed policy can be used with high confidence in its future performance. A simple principled criteria for identifying when training has completed is when the error bounds on the value estimates of the current policy are sufficiently small that the optimal policy is fixed, with high probability. We present an upper bound on the amount of training data required to identify the optimal policy as a function of the unknown separation gap between the optimal and the next-best policy values. We illustrate with several small problems that by estimating this gap in an online manner, the number of training samples to provably reach optimality can be significantly lower than predicted offline using a Probably Approximately Correct framework that requires an input epsilon parameter.

We present a novel framework for learning to interpret and generate language using only perceptual context as supervision. We demonstrate its capabilities by developing a system that learns to sportscast simulated robot soccer games in both English and Korean without any language-specific prior knowledge. Training employs only ambiguous supervision consisting of a stream of descriptive textual comments and a sequence of events extracted from the simulation trace. The system simultaneously establishes correspondences between individual comments and the events that they describe while building a translation model that supports both parsing and generation. We also present a novel algorithm for learning which events are worth describing. Human evaluations of the generated commentaries indicate they are of reasonable quality and in some cases even on par with those produced by humans for our limited domain.

Decentralized planning in uncertain environments is a complex task generally dealt with by using a decision-theoretic approach, mainly through the framework of Decentralized Partially Observable Markov Decision Processes (DEC-POMDPs). Although DEC-POMDPS are a general and powerful modeling tool, solving them is a task with an overwhelming complexity that can be doubly exponential. In this paper, we study an alternate formulation of DEC-POMDPs relying on a sequence-form representation of policies. From this formulation, we show how to derive Mixed Integer Linear Programming (MILP) problems that, once solved, give exact optimal solutions to the DEC-POMDPs. We show that these MILPs can be derived either by using some combinatorial characteristics of the optimal solutions of the DEC-POMDPs or by using concepts borrowed from game theory. Through an experimental validation on classical test problems from the DEC-POMDP literature, we compare our approach to existing algorithms. Results show that mathematical programming outperforms dynamic programming but is less efficient than forward search, except for some particular problems. The main contributions of this work are the use of mathematical programming for DEC-POMDPs and a better understanding of DEC-POMDPs and of their solutions. Besides, we argue that our alternate representation of DEC-POMDPs could be helpful for designing novel algorithms looking for approximate solutions to DEC-POMDPs.