Recent studies in the field of human vision science suggest that the human responses to the stimuli on a visual display are non-deterministic. People may attend to different locations on the same visual input at the same time. Based on this knowledge, we propose a new stochastic model of visual attention by introducing a dynamic Bayesian network to predict the likelihood of where humans typically focus on a video scene. The proposed model is composed of a dynamic Bayesian network with 4 layers. Our model provides a framework that simulates and combines the visual saliency response and the cognitive state of a person to estimate the most probable attended regions. Sample-based inference with Markov chain Monte-Carlo based particle filter and stream processing with multi-core processors enable us to estimate human visual attention in near real time. Experimental results have demonstrated that our model performs significantly better in predicting human visual attention compared to the previous deterministic models.

We develop multiattribute auctions that accommodate generalized additive independent (GAI) preferences. We propose an iterative auction mechanism that maintains prices on potentially overlapping GAI clusters of attributes, thus decreases elicitation and computational burden, and creates an open competition among suppliers over a multidimensional domain. Most significantly, the auction is guaranteed to achieve surplus which approximates optimal welfare up to a small additive factor, under reasonable equilibrium strategies of traders. The main departure of GAI auctions from previous literature is to accommodate non-additive trader preferences, hence allowing traders to condition their evaluation of specific attributes on the value of other attributes. At the same time, the GAI structure supports a compact representation of prices, enabling a tractable auction process. We perform a simulation study, demonstrating and quantifying the significant efficiency advantage of more expressive preference modeling. We draw random GAI-structured utility functions with various internal structures, generate additive functions that approximate the GAI utility, and compare the performance of the auctions using the two representations. We find that allowing traders to express existing dependencies among attributes improves the economic efficiency of multiattribute auctions.

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.

LEXSYS, (Legume Expert System) was a project conceived at IITA (International Institute of Tropical Agriculture) Ibadan Nigeria. It was initiated by the COMBS (Collaborative Group on Maize-Based Systems Research in the 1990. It was meant for a general framework for characterizing on-farm testing for technology design for sustainable cereal-based cropping system. LEXSYS is not a true expert system as the name would imply, but simply a user-friendly information system. This work is an attempt to give a formal representation of the existing system and then present areas where intelligent agent can be applied.

Probabilistic matrix factorization (PMF) is a powerful method for modeling data associated with pairwise relationships, finding use in collaborative filtering, computational biology, and document analysis, among other areas. In many domains, there is additional information that can assist in prediction. For example, when modeling movie ratings, we might know when the rating occurred, where the user lives, or what actors appear in the movie. It is difficult, however, to incorporate this side information into the PMF model. We propose a framework for incorporating side information by coupling together multiple PMF problems via Gaussian process priors. We replace scalar latent features with functions that vary over the space of side information. The GP priors on these functions require them to vary smoothly and share information. We successfully use this new method to predict the scores of professional basketball games, where side information about the venue and date of the game are relevant for the outcome.

Current statistical models for structured prediction make simplifying assumptions about the underlying output graph structure, such as assuming a low-order Markov chain, because exact inference becomes intractable as the tree-width of the underlying graph increases. Approximate inference algorithms, on the other hand, force one to trade off representational power with computational efficiency. In this paper, we propose two new types of probabilistic graphical models, large margin Boltzmann machines (LMBMs) and large margin sigmoid belief networks (LMSBNs), for structured prediction. LMSBNs in particular allow a very fast inference algorithm for arbitrary graph structures that runs in polynomial time with a high probability. This probability is data-distribution dependent and is maximized in learning. The new approach overcomes the representation-efficiency trade-off in previous models and allows fast structured prediction with complicated graph structures. We present results from applying a fully connected model to multi-label scene classification and demonstrate that the proposed approach can yield significant performance gains over current state-of-the-art methods.

In this paper, we describe our work on discovering missing links in Wikipedia articles. This task is important for both readers and authors of Wikipedia. The readers will benefit from the increased article quality with better navigation support. On the other hand, the system can be employed to support the authors during editing. This study combines the strengths of different approaches previously applied for the task, and adds its own techniques to reach satisfactory results. Because of the subjectivity in the nature of the task; automatic evaluation is hard to apply. Comparing approaches seems to be the best method to evaluate new techniques, and we offer a semi-automatized method for evaluation of the results. The recall is calculated automatically using existing links in Wikipedia. The precision is calculated according to manual evaluations of human assessors. Comparative results for different techniques are presented, showing the success of our improvements. We employ Turkish Wikipedia, we are the first to study on it, to examine whether a small instance is scalable enough for such purposes.

Given the prevalence of data leaks, organizations appreciate the importance of implementing privacy policies to protect sensitive data. The growing field of Data Loss Prevention (DLP) offers tools to enforce such policies for both data stored within an organization and data being shared outside of an organization (e.g. through email). While the DLP community has given much attention to the problem of enforcing data privacy policies in a comprehensive manner, little has been done to support the development of such policies. We present a small user study demonstrating that developing such policies is also a very challenging problem. In our study, users were asked to evaluate various expressive file names for sensitivity; that it, they were asked to consider how broadly they were willing to share those filenames both inside and outside their place of employment. The study indicates that users interpret their employer’s privacy concerns in differing ways, resulting in complex, personalized privacy policies at the user end. These results suggest that it may be difficult for users to form a coherent organization-level privacy policy and that the results of a DLP-based enforcement of such policies (e.g. quarantined emails) may be confusing for many users in the organization.

While there is a plethora of mechanisms to ensure lawful access to privacy-protected data, additional research is required in order to reassure individuals that their personal data is being used for the purpose that they consented to. This is particularly important in the context of new data mining approaches, as used, for instance, in biomedical research and commercial data mining. We argue for the use of computational workflows to ensure and enforce appropriate use of sensitive personal data. Computational workflows describe in a declarative manner the data processing steps and the expected results of complex data analysis processes such as data mining (Gil et al. 2007b; Taylor et al. 2006). We see workflows as an artifact that captures, among other things, how data is being used and for what purpose. Existing frameworks for computational workflows need to be extended to incorporate privacy policies that can govern the use of data.