Learners may develop expertise by experiencing numerous different but relevant situations. Computer games and virtual simulations can facilitate these training opportunities, however, because of the relative difficulty in authoring new scenarios, the increasing need for new and different scenarios becomes a bottleneck in the learning process. Furthermore, a one-size-fits-all scenario may not address all of the abilities, needs, or goals of a particular learner. To address these issues we present a novel technique, Automated Scenario Adaptation, to automatically “rewrite” narrative scenario content to suit individual learners’ needs and abilities and to incorporate recent changes from real world learning needs. Scenario adaptation acts as problem generation for intelligent tutoring systems, producing greater learning opportunities that facilitate engagement and continued learner involvement.

The Snackbot, is a robot designed in collaboration between the Robotics Institute, and the Human Computer Interaction Institute of Carnegie Mellon University. The Snackbot was created to traverse the halls of Carnegie Mellon University, and deliver food items ordered by occupants of the offices. The goal of this development project for the Snackbot, was to refine the audio/video synchronization, and to also create a simple way to log, and stream that data over a network. Such a task requires that one not only carefully consider different pieces of software to use, but also that they can apply it across the necessary platform. For the Snackbot, the sight, and sound are important qualities, especially when testing out in the field using an operator. That ability is crucial when preparing an interactive robot to autonomously carry out its task efficiently.

This paper tackles the problem of building situated prompting and assistance systems for guiding a human with a cognitive disability through a large domain containing multiple tasks. This problem is challenging because the target population has difficulty maintaining goals, recalling necessary steps and recognizing objects and potential actions (affordances), and therefore may not appear to be acting rationally. Prompts or cues from an automated system can be very helpful in this regard, but the domain is inherently partially observable due to sensor noise and uncertain human behaviours, making the task of selecting an appropriate prompt very challenging. Prior work has shown how such automated assistance for a single task can be modeled as a partially observable Markov decision process (POMDP). In this paper, we generalise this to multiple tasks, and show how to build a scalable, distributed and hierarchical controller. We demonstrate the algorithm in a set of simulated domains and show it can perform as well as the full model in many cases, and can give solutions to large problems (over 10 15 states and 10 9 observations) for which the full model fails to find a policy.

The New York Public Library is participating in the Chronicling America initiative to develop an online searchable database of historically significant newspaper articles. Microfilm copies of the papers are scanned and high resolution OCR software is run on them. The text from the OCR provides a wealth of data and opinion for researchers and historians. However, the categorization of articles provided by the OCR engine is rudimentary and a large number of the articles are labeled ``editorial" without further categorization. To provide a more refined grouping of articles, unsupervised machine learning algorithms (such as K-Means) are being investigated. The K-Means algorithm requires tuning of parameters such as the number of clusters and mechanism of seeding to ensure that the search is not prone to being caught in a local minima. We designed a pilot study to observe whether humans are adept at finding sub-categories. The subjective labels provided by humans are used as a guide to compare performance of the automated clustering techniques. In addition, seeds provided by annotators are carefully incorporated into a semi-supervised K-Means algorithm (Seeded K-Means); empirical results indicate that this helps to improve performance and provides an intuitive sub-categorization of the articles labeled ``editorial" by the OCR engine.

We present an efficient random decision tree algorithm for case-based reasoning systems. We combine this algorithm with a simple similarity measure based on domain knowledge to create a stronger hybrid algorithm. This combination is based on our general approach for combining lazy and eager learning methods. We evaluate the resulting algorithms on a case base of patient records in a palliative care domain. Our hybrid algorithm consistently produces a lower average error than the base algorithms.

Historically, causal markers, syntactic structures and connectives have been the sole identifying features for automatically extracting causal relations in natural language discourse. However various connectives such as “and,” prepositions such as “as” and other syntactic structures are highly ambiguous in nature, and it is clear that one cannot solely rely on lexico-syntactic markers for detection of causal phenomenon in discourse. This paper introduces the theory of granularity and describes different approaches to identify granularity in natural language. As causality is often granular in nature, we use granularity relations to discover and infer the presence of causal relations in text. We compare this with causal relations identified using just causal markers. We achieve a precision of 0.91 and a recall of 0.79 using granularity for causal relation detection, as compared to a precision of 0.79 and a recall of 0.44 using pure causal markers for causality detection.

We propose a new interpretation for a certain kind of ASO preference rules to handling prioritized symptoms. Our interpretation is based on the idea of dependence based priority propagation. Then, we present a fixpoint based answer set optimization method under such an interpretation.

The Upper Confidence bounds for Trees (UCT) algorithm has in recent years captured the attention of the planning and game-playing community due to its notable success in the game of Go. However, attempts to reproduce similar levels of performance in domains that are the forte of Minimax-style algorithms have been largely unsuccessful, making any comparative studies of the two hard. In this paper, we study UCT in the game of Mancala, which to our knowledge is the first domain where both search algorithms perform quite well with minimal enhancement. We focus on the three key components of the UCT algorithm in its purest form - targeted node expansion, state value estimation via playouts and averaging backups - and look at their contributions to the overall performance of the algorithm. We study the trade-offs involved in using alternate ways to perform these steps. Finally, we demonstrate a novel hybrid approach to enhancing UCT, that exploits its superior decision accuracy in regions of the search space with few terminal nodes.

In this paper, we present an image and text mining tool named TNT. This tool is based on Contextual Exploration and work on different points of view. It can process a corpus of all sizes in French or in English. The web interface associated with this tool, offers a reorganization of the text guided by the images and annotated segments that are associated.

We consider the problem of planning, on the fly, a path whereby a robotic vehicle will cover every point in an initially unknown spatial environment. We describe four strategies (Iterated WaveFront, Greedy-Scan, Delayed Greedy-Scan and Closest-First Scan) for generating cost-effective coverage plans in real time for unknown environments. We give theorems showing the correctness of our planning strategies. Our experiments demonstrate that some of these strategies work significantly better than others, and that the best ones work very well; e.g., in environments having an average of 64,000 locations for the robot to cover, the best strategy returned plans with less than 6% redundant coverage, and took only an average of 0.1 milliseconds per action.