In multiple-Instance Learning (MIL), training class labels are attached to sets of bags composed of unlabeled instances, and the goal is to deal with classification of bags. Most previous MIL algorithms, which tackle classification problems, consider each instance as a represented feature. Although the algorithms work well in some prediction problems, considering diverse features to represent an instance may provide more significant information for learning task. Moreover, since each instance may be mapped into diverse feature spaces, encountering a large number of irrelevant or redundant features is inevitable. In this paper, we propose a method to select relevant instances and concurrently consider multiple features for each instance, which is termed as MIL-MFS. MIL-MFS is based on multiple kernel learning (MKL), and it iteratively selects the fusing multiple features for classifier training. Experimental results show that the MIL-MFS combined with multiple kernel learning can significantly improve the classification performance.

Multilabel classification is a central problem in many areas of data analysis, including text and multimedia categorization, where individual data objects need to be assigned multiple labels. A key challenge in these tasks is to learn a classifier that can properly exploit label correlations without requiring exponential enumeration of label subsets during training or testing. We investigate novel loss functions for multilabel training within a large margin framework---identifying a simple alternative that yields improved generalization while still allowing efficient training. We furthermore show how covariances between the label models can be learned simultaneously with the classification model itself, in a jointly convex formulation, without compromising scalability. The resulting combination yields state of the art accuracy in multilabel webpage classification.

Social influence has become the essential factor which drives the dynamic evolution process of social network structure and user behaviors. Previous research often focus on social influence analysis in network-level or topic-level. In this paper, we concentrate on predicting item-level social influence to reveal the users' influences in a more fine-grained level. We formulate the social influence prediction problem as the estimation of a user-post matrix, where each entry in the matrix represents the social influence strength the corresponding user has given the corresponding web post. To deal with the sparsity and complex factor challenges in the research, we model the problem by extending the probabilistic matrix factorization method to incorporate rich prior knowledge on both user dimension and web post dimension, and propose the Probabilistic Hybrid Factor Matrix Factorization (PHF-MF) approach. Intensive experiments are conducted on a real world online social network to demonstrate the advantages and characteristics of the proposed method.

In this paper, we extend the progression semantics for first-order disjunctive logic programs and show that it coincides with the stable model semantics. Based on it, we further show how disjunctive answer set programming is related to Satisfiability Modulo Theories.

This paper explores a new measure of similarity between point sets in arbitrary metric spaces. The measure is based on the spatial overlap of the “shapes” and “densities” of these point sets. It is applicable in any domain where point sets are a natural representation for data. Specifically, we show examples of its use in natural language processing, object recognition in images and point set classification. We provide a geometric interpretation of this measure and show that it is well-motivated, intuitive, parameter-free, and straightforward to use. We further demonstrate that it is computationally tractable and applicable to both supervised and unsupervised learning problems.

Possessing a sufficient level of situation awareness is essential for effective decision making in dynamic environments. In video games, this includes being aware to some extent of the intentions of the opponents. Such high-level awareness hinges upon inferences over the lower-level situation awareness provided by the game state. Traditional plan recognizers are completely passive processes that leave all the initiative to the observed agent. In a situation where the opponent's intentions are unclear, the observer is forced to wait until further observations of the opponent's actions are made to disambiguate the pending goal hypotheses. With the plan recognizer we propose, in contrast, the observer would take the initiative and provoke the opponent, with the expectation that his reaction will give cues as to what his true intentions actually are.

In this paper, we model the pair-wise similarities of a setof documents as a weighted network with a single cutoffparameter. Such a network can be thought of an ensemble of unweighted graphs, each consisting of edges withweights greater than the cutoff value. We look at this network ensemble as a complex system with a temperature parameter, and refer to it as a Latent Network. Ourexperiments on a number of datasets from two different domains show that certain properties of latent networks like clustering coefficient, average shortest path,and connected components exhibit patterns that are significantly divergent from randomized networks. We explain that these patterns reflect the network phase transition as well as the existence of a community structure in document collections. Using numerical analysis,we show that we can use the aforementioned networkproperties to predicts the clustering Normalized MutualInformation (NMI) with high correlation (rho > 0.9). Finally we show that our clustering method significantlyoutperforms other baseline methods (NMI > 0.5)

Automatic semantic annotation of structured data enables unsupervised integration of data from heterogeneous sources but is difficult to perform accurately due to the presence of many numeric fields and proper-noun fields that do not allow reference-based approaches and the absence of natural language text that prevents the use of language-based approaches. In addition, several of these semantic types have multiple heterogeneous representations, while sharing syntactic structure with other types. In this work, we propose a new approach to use conditional random fields (CRFs) to perform semantic annotation of structured data that takes advantage of the structure and labels of the tokens for higher accuracy of field labeling, while still allowing the use of exact inference techniques. We compare our approach with a linear-CRF based model that only labels fields and also with a regular-expression based approach.

In constraint solving, a critical bottleneck is the formulation of aneffective constraint model of an input problem. The Conjure system describedin this paper, a substantial step forward over prototype versions of Conjurepreviously reported, makes a valuable contribution to the automation ofconstraint modelling by automatically producing constraint models from theirspecifications in the abstract constraint specification language Essence. Aset of rules is used to refine an abstract specification into a concreteconstraint model. We demonstrate that this set of rules is readily extensibleto increase the space of possible constraint models Conjure can produce. Ourempirical results confirm that Conjure can reproduce successfully the kernelsof the constraint models of 32 benchmark problems found in the literature.

We study probabilistic variants of the description logic EL. For the case where probabilities apply only to concepts, we provide a careful analysis of the borderline between tractability and ExpTime-completeness. One outcome is that any probability value except zero and one leads to intractability in the presence of general TBoxes, while this is not the case for classical TBoxes. For the case where probabilities can also be applied to roles, we show PSpace-completeness. This result is (positively) surprising as the best previously known upper bound was 2-ExpTime and there were reasons to believe in completeness for this class.