The MAP (maximum a posteriori assignment) problem in Bayesian networks is the problem of finding the most probable instantiation of a set of variables given partial evidence for the remaining variables. The state-of-the-art exact solution method is depth-first branch-and-bound search using dynamic variable ordering and a jointree upper bound proposed by Park and Darwiche [2003]. Since almost all search time is spent computing the jointree bounds, we introduce an efficient method for computing these bounds incrementally. We point out that, using a static variable ordering, it is only necessary to compute relevant upper bounds at each search step, and it is also possible to cache potentials of the jointree for efficient backtracking. Since the jointree computation typically produces bounds for joint configurations of groups of variables, our method also instantiates multiple variables at each search step, instead of a single variable, in order to reduce the number of times that upper bounds need to be computed. Experiments show that this approach leads to orders of magnitude reduction in search time.

Unsupervised learning of morphology is an important task for human learners and in natural language processing systems. Previous systems focus on segmenting words into substrings (taking ⇒ tak.ing), but sometimes a segmentation-only analysis is insufficient (e.g., taking may be more appropriately analyzed as take+ing, with a spelling rule accounting for the deletion of the stem-final e). In this paper, we develop a Bayesian model for simultaneously inducing both morphology and spelling rules. We show that the addition of spelling rules improves performance over the baseline morphology-only model.

Gaussian Processes (GPs) are promising Bayesian methods for classification and regression problems. They have also been used for semi-supervised learning tasks. In this paper, we propose a new algorithm for solving semi-supervised binary classification problem using sparse GP regression (GPR) models. It is closely related to semi-supervised learning based on support vector regression (SVR) and maximum margin clustering. The proposed algorithm is simple and easy to implement. It gives a sparse solution directly unlike the SVR based algorithm. Also, the hyperparameters are estimated easily without resorting to expensive cross-validation technique. Use of sparse GPR model helps in making the proposed algorithm scalable. Preliminary results on synthetic and real-world data sets demonstrate the efficacy of the new algorithm.

Can we devise educational systems that provide individualized instruction tailored to the needs of the individual learners, as many good teachers do? Intelligent Tutoring Systems is the interdisciplinary field that investigates this question by integrating research in Artificial Intelligence, Cognitive Science and Education. Research in this field has successfully delivered techniques and systems that provide adaptive support for student problem solving in variety of domains. There are, however, other educational activities that can benefit from individualized computer-based support, such as studying examples, exploring interactive simulations and playing educational games. Providing individualized support for these activities rises unique challenges, because it requires that an ITS can model and adapt to student behaviors, skills and mental states often not as structured and well-defined as those involved in traditional problem solving. I will present a variety of projects that illustrate some of these challenges, our proposed solutions, and future opportunities.

Mechanical design tools would be considerably more useful if we could interact with them in the way that human designers communicate design ideas to one another, i.e., using crude sketches and informal speech. Those crude sketches frequently contain pen strokes of two different sorts, one type portraying device structure, the other denoting gestures, such as arrows used to indicate motion. We report here on techniques we developed that use information from both sketch and speech to distinguish gesture strokes from non-gestures -- a critical first step in understanding a sketch of a device. We collected and analyzed unconstrained device descriptions, which revealed six common types of gestures. Guided by this knowledge, we developed a classifier that uses both sketch and speech features to distinguish gesture strokes from non-gestures. Experiments with our techniques indicate that the sketch and speech modalities alone produce equivalent classification accuracy, but combining them produces higher accuracy.

We present an approach to semi-supervised learning based on an exponential family characterization. Our approach generalizes previous work on coupled priors for hybrid generative/discriminative models. Our model is more flexible and natural than previous approaches. Experimental results on several data sets show that our approach also performs better in practice. 

Logic reasoning approaches to fault diagnosis account for the fact that a component c j may fail intermittently by introducing a parameter g j that expresses the probability the component exhibits correct behavior. This component parameter g j , in conjunction with a priori fault probability, is usedin a Bayesian framework to compute the posterior fault candidate probabilities. Usually, information on g j is not known a priori. While proper estimation of g j can have a great impact on the diagnostic accuracy, at present, only approximations have been proposed. We present a novel framework, BARINEL, that computes exact estimations of g j as integral part of the posterior candidate probability computation. BARINEL’s diagnostic performance is evaluated for both synthetic and real software systems. Our results show that our approach is superior to approaches based on classical persistent fault models as well as previously proposed intermittent fault models.

Event detection is a critical task in sensor networks for a variety of real-world applications. Many real-world events often exhibit complex spatio-temporal patterns whereby they manifest themselves via observations over time and space proximities. These spatio-temporal events cannot be handled well by many of the previous approaches. In this paper, we propose a new Spatio-Temporal Event Detection (STED) algorithm in sensor networks based on a dynamic conditional random field (DCRF) model. Our STED method handles the uncertainty of sensor data explicitly and permits neighborhood interactions in both observations and event labels. Experiments on both real data and synthetic data demonstrate that our STED method can provide accurate event detection in near real time even for large-scale sensor networks.

This paper deals with Decision-Making in the context of multiattribute utility theory and, more precisely, with the problem of efficiently determining the best alternative w.r.t. an agent's preferences (choice problem). We assume that alternatives are elements of a product set of attributes and that the agent's preferences are represented by a generalized additive decomposable (GAI) utility on this set. Such a function allows an efficient representation of interactions between attributes while preserving some decomposability of the model. GAI utilities can be compiled into graphical structures called GAI networks that can be exploited to solve choice problems using collect/distribute schemes essentially similar to those used in Bayesian networks. In this paper, rather than directly using this scheme on the GAI network for determining the most preferred alternative, we propose to work with another GAI function, acting as an upper-bound on utility values and enhancing the model's decomposability. This method still provides the exact optimal solution but speeds up significantly the search. It proves to be particularly useful when dealing with choice and ranking under constraints and within collective Decision-Making, where GAI nets tend to have a large size. We present an efficient algorithm for determining this new GAI function and provide experimental results highlighting the practical efficiency of our procedure.

Users share and access large volumes of information on social networking sites like Facebook, Flickr, del.icio.us, etc. Whereas a few of these sites have generic, impersonal searching mechanisms, we have developed an agent-based framework that mines the social network of a user to improve search results. Our Social Network-based Item Search (SNIS) system uses agents that utilize the connections of a user in the social network to facilitate the search for items of interest. Our approach generates targeted search results that can improve the precision of the result returned from a user's query. We have implemented the SNIS agent-based framework in Flickr, a photo-sharing social network, for searching for photos by using tag lists as search queries. We discuss the architecture of SNIS, motivate the searching scheme used, and demonstrate the effectiveness of the SNIS approach by presenting results. We also show how SNIS can be utilized for expertise location.