The Green Driver app is a dynamic routing application for GPS-enabled smartphones. Green Driver combines client GPS data with real-time traffic light information provided by cities to determine optimal routes in response to driver route requests. Routes are optimized with respect to travel time, with the intention of saving the driver both time and fuel, and rerouting can occur if warranted. During a routing session, client phones communicate with a centralized server that both collects GPS data and processes route requests. All relevant data are anonymized and saved to databases for analysis; statistics are calculated from the aggregate data and fed back to the routing engine to improve future routing. Analyses can also be performed to discern driver trends: where do drivers tend to go, how long do they stay, when and where does traffic congestion occur, and so on. The system uses a number of techniques from the field of artificial intelligence. We apply a variant of A* search for solving the stochastic shortest path problem in order to find optimal driving routes through a network of roads given light-status information. We also use dynamic programming and hidden Markov models to determine the progress of a driver through a network of roads from GPS data and light-status data. The Green Driver system is currently deployed for testing in Eugene, Oregon, and is scheduled for large-scale deployment in Portland, Oregon, in Spring 2011.

We present a new approach to characterizing the semantics for the integration of rules and first-order logic in general, and description logics in particular, based on a circumscription characterization of answer set programming, introduced earlier by Lin and Zhou. We show that both Rosati's semantics based on NM-models and Lukasiewicz's answer set semantics can be characterized by circumscription, and the difference between the two can be seen as a matter of circumscription policies. This approach leads to a number of new insights. First, we rebut a criticism on Lukasiewicz's semantics for its inability to reason for negative consequences. Second, our approach leads to a spectrum of possible semantics based on different circumscription policies, and shows a clear picture of how they are related. Finally, we show that the idea of this paper can be applied to first-order general stable models.

Latent relational search (LRS) is a novel approach for mapping knowledge across two domains. Given a source domain knowledge concerning the Moon, "The Moon is a satellite of the Earth," one can form a question {(Moon, Earth), (Ganymede, ?)} to query an LRS engine for new knowledge in the target domain concerning the Ganymede. An LRS engine relies on some supporting sentences such as ``Ganymede is a natural satellite of Jupiter.'' to retrieve and rank "Jupiter" as the first answer. This paper proposes cross-language latent relational search (CLRS) to extend the knowledge mapping capability of LRS from cross-domain knowledge mapping to cross-domain and cross-language knowledge mapping. In CLRS, the supporting sentences for the source pair might be in a different language with that of the target pair. We represent the relation between two entities in an entity pair by lexical patterns of the context surrounding the two entities. We then propose a novel hybrid lexical pattern clustering algorithm to capture the semantic similarity between paraphrased lexical patterns across languages. Experiments on Japanese-English datasets show that the proposed method achieves an MRR of 0.579 for CLRS task, which is comparable to the MRR of an existing monolingual LRS engine.

We consider a learning setting of importance to large scale machine learning: potentially unlimited data arrives sequentially, but only a small fraction of it is labeled. The learner cannot store the data; it should learn from both labeled and unlabeled data, and it may also request labels for some of the unlabeled items. This setting is frequently encountered in real-world applications and has the characteristics of online, semi-supervised, and active learning. Yet previous learning models fail to consider these characteristics jointly. We present OASIS, a Bayesian model for this learning setting. The main contributions of the model include the novel integration of a semi-supervised likelihood function, a sequential Monte Carlo scheme for efficient online Bayesian updating, and a posterior-reduction criterion for active learning. Encouraging results on both synthetic and real-world optical character recognition data demonstrate the synergy of these characteristics in OASIS.

Web Service (WS) domains constitute an application field where automated planning can significantly contribute towards achieving customisable and adaptable compositions. Following the vision of using domain-independent planning and declarative complex goals to generate compositions based on atomic service descriptions, we apply a planning framework based on Constraint Satisfaction techniques to a domain consisting of WSs with diverse functionalities. One of the key requirements of such domains is the ability to address the incomplete knowledge problem, as well as recovering from failures that may occur during execution. We propose an algorithm for interleaving planning, monitoring and execution, where continual planning via altering the CSP is performed, under the light of the feedback acquired at runtime. The system is evaluated against a number of scenarios including real WSs, demonstrating the leverage of situations that can be effectively tackled with respect to previous approaches.

We present an algorithm for compressing pattern databases (PDBs) and a method for fast random access of these com-pressed PDBs. We demonstrate the effectiveness of our technique by compressing two 6-tile sliding-tile PDBs by a factor of 12 and a 7-tile sliding-tile PDB by a factor of 24.

Suppose an unpredictable evader is free to move around in a polygonal environment of arbitrary complexity that is under full camera surveillance. How many pursuers, each with the same maximum speed as the evader, are necessary and sufficient to guarantee a successful capture of the evader? The pursuers always know the evader's current position through the camera network, but need to physically reach the evader to capture it. We allow the evader the knowledge of the current positions of all the pursuers as well---this accords with the standard worst-case analysis model, but also models a practical situation where the evader has ``hacked'' into the surveillance system. Our main result is to prove that three pursuers are always sufficient and sometimes necessary to capture the evader. The bound is independent of the number of vertices or holes in the polygonal environment.

Collaborative tagging systems are now deployed extensivelyto help users share and organize resources.Tag prediction and recommendation systems generallymodel user behavior as research has shown that accuracycan be significantly improved by modeling users’preferences. However, these preferences are usuallytreated as constant over time, neglecting the temporalfactor within users’ interests. On the other hand, littleis known about how this factor may influence predictionin social bookmarking systems. In this paper, weinvestigate the temporal dynamics of user interests intagging systems and propose a user-tag-specific temporalinterests model for tracking users’ interests overtime. Additionally, we analyze the phenomenon of topicswitches in social bookmarking systems, showing that atemporal interests model can benefit from the integrationof topic switch detection and that temporal characteristicsof social tagging systems are different fromtraditional concept drift problems. We conduct experimentson three public datasets, demonstrating the importanceof personalization and user-tag specializationin tagging systems. Experimental results show that ourmethod can outperform state-of-the-art tag predictionalgorithms. We also incorporate our model within existingcontent-based methods yielding significant improvementsin performance.

We address the problem of building a continuous occupancy representation of the environment with ranging sensors. Observations from such sensors provide two types of information: a line segment or a beam indicating no returns along them (free-space); a point or return at the end of the segment representing an occupied surface. To model these two types of observations in a principled statistical manner, we propose a novel methodology based on integral kernels. We show that integral kernels can be directly incorporated into a Gaussian process classification (GPC) framework to provide a continuous non-parametric Bayesian estimation of occupancy. Directly handling line segment and point observations avoids the need to discretise segments into points, reducing the computational cost of GPC inference and learning. We present experiments on 2D and 3D datasets demonstrating the benefits of the approach.

Crowdsourcing platforms such as Amazon Mechanical Turk have become popular for a wide variety of human intelligence tasks; however, quality control continues to be a significant challenge. Recently, we propose TurKontrol, a theoretical model based on POMDPs to optimize iterative, crowd-sourced workflows. However, they neither describe how to learn the model parameters, nor show its effectiveness in a real crowd-sourced setting. Learning is challenging due to the scale of the model and noisy data: there are hundreds of thousands of workers with high-variance abilities. This paper presents an end-to-end system that first learns TurKontrol's POMDP parameters from real Mechanical Turk data, and then applies the model to dynamically optimize live tasks. We validate the model and use it to control a successive-improvement process on Mechanical Turk. By modeling worker accuracy and voting patterns, our system produces significantly superior artifacts compared to those generated through nonadaptive workflows using the same amount of money.