In this paper, we propose a new framework for constructing text metrics which can be used to compare and support inferences among terms and sets of terms. Our metric is derived from data-driven kernels on graphs that let us capture global relations among terms and sets of terms, regardless of their complexity and size. To compute the metric efficiently for any two subsets of terms, we develop an approximation technique that relies on the precompiled term-term similarities. To scale-up the approach to problems with huge number of terms, we develop and experiment with a solution that subsamples the term space. We demonstrate the benefits of the whole framework on two text inference tasks: prediction of terms in the article from its abstract and query expansion in information retrieval.

Activity recognition using mobile phones has great potential in many applications including mobile healthcare. In order to let a person easily know whether he is in strict compliance with the doctor's exercise prescription and adjust his exercise amount accordingly, we can use a smart-phone based activity reporting system to accurately recognize a range of daily activities and report the duration of each activity. A triaxial accelerometer embedded in the smart phone is used for the classification of several activities, such as staying still, walking, running, and going upstairs and downstairs. The model learnt from a specific person often cannot yield accurate results when used on a different person. To solve the cross-people activity recognition problem, we propose an algorithm known as TransEMDT (Transfer learning EMbedded Decision Tree) that integrates a decision tree and the k-means clustering algorithm for personalized activity-recognition model adaptation. Tested on a real-world data set, the results show that our algorithm outperforms several traditional baseline algorithms.

Matrix factorization based techniques, such as nonnegative matrix factorization (NMF) and concept factorization (CF), have attracted great attention in dimension reduction and data clustering. Both of them are linear learning problems and lead to a sparse representation of the data. However, the sparsity obtained by these methods does not always satisfy locality conditions, thus the obtained data representation is not the best. This paper introduces a locality-constrained concept factorization method which imposes a locality constraint onto the traditional concept factorization. By requiring the concepts (basis vectors) to be as close to the original data points as possible, each data can be represented by a linear combination of only a few basis concepts. Thus our method is able to achieve sparsity and locality at the same time. We demonstrate the effectiveness of this novel algorithm through a set of evaluations on real world applications.

Many AI applications are based on some underlying logic that tolerates inconsistent information in a non-trivial way. However, it is not always clear what should be the exact nature of such a logic, and how to choose one for a specific application. In this paper, we formulate a list of desirable properties of `ideal' logics for reasoning with inconsistency, identify a variety of logics that have these properties, and provide a systematic way of constructing, for every n>2, a family of such n-valued logics.

In this paper, we propose to extract a compact yet discriminative visual descriptor directly on the mobile device, which tackles the wireless query transmission latency in mobile landmark search. This descriptor is offline learnt from the location contexts of geo-tagged Web photos from both Flickr and Panoramio with two phrases: First, we segment the landmark photo collections into discrete geographical regions using a Gaussian Mixture Model [Stauffer et al., 2000]. Second, a ranking sensitive vocabulary boosting is introduced to learn a compact codebook within each region. To tackle the locally optimal descriptor learning caused by imprecise geographical segmentation, we further iterate above phrases by feedback an “entropy” based descriptor compactness into a prior distribution to constrain the Gaussian mixture modeling. Consequently, when entering a specific geographical region, the codebook in the mobile device is downstream adapted, which ensures efficient extraction of compact descriptor, its low bit rate transmission, as well as promising discrimination ability. We deploy our descriptor within both HTC and iPhone mobile phones, testing landmark search in typical areas included Beijing, New York, and Barcelona containing one million images. Our learning descriptor outperforms alternative compact descriptors [Chen et al., 2009][Chen et al., 2010][Chandrasekhar et al., 2009a][Chandrasekhar et al., 2009b] with a large margin.

Understanding the rapidly growing short text is very important. Short text is different from traditional documents in its shortness and sparsity, which hinders the application of conventional machine learning and text mining algorithms. Two major approaches have been exploited to enrich the representation of short text. One is to fetch contextual information of a short text to directly add more text; the other is to derive latent topics from existing large corpus, which are used as features to enrich the representation of short text. The latter approach is elegant and efficient in most cases. The major trend along this direction is to derive latent topics of certain granularity through well-known topic models such as latent Dirichlet allocation (LDA). However, topics of certain granularity are usually not sufficient to set up effective feature spaces. In this paper, we move forward along this direction by proposing an method to leverage topics at multiple granularity, which can model the short text more precisely. Taking short text classification as an example, we compared our proposed method with the state-of-the-art baseline over one open data set. Our method reduced the classification error by 20.25% and 16.68%respectively on two classifiers.

We develop a real-time algorithm based on a Monte-Carlo game tree search for solving a quantified constraint satisfaction problem (QCSP), which is a CSP where some variables are universally quantified. A universally quantified variable represents a choice of nature or an adversary. The goal of a QCSP is to make a robust plan against an adversary. However, obtaining a complete plan off-line is intractable when the size of the problem becomes large. Thus, we need to develop a real-time algorithm that sequentially selects a promising value at each deadline. Such a problem has been considered in the field of game tree search. In a standard game tree search algorithm, developing a good static evaluation function is crucial. However, developing a good static evaluation function for a QCSP is very difficult since it must estimate the possibility that a partially assigned QCSP is solvable. Thus, we apply a Monte-Carlo game tree search technique called UCT. However, the simple application of the UCT algorithm does not work since the player and the adversary are asymmetric, i.e., finding a game sequence where the player wins is very rare. We overcome this difficulty by introducing constraint propagation techniques. We experimentally compare the winning probability of our UCT-based algorithm and the state-of-the-art alpha-beta search algorithm. Our results show that our algorithm outperforms the state-of-the-art algorithm in large-scale problems.

Multi-label learning deals with the problem where each training example is represented by a single instance while associated with a set of class labels. For an unseen example, existing approaches choose to determine the membership of each possible class label to it based on identical feature set, i.e. the very instance representation of the unseen example is employed in the discrimination processes of all labels. However, this commonly-used strategy might be suboptimal as different class labels usually carry specific characteristics of their own, and it could be beneficial to exploit different feature sets for the discrimination of different labels. Based on the above reflection, we propose a new strategy to multi-label learning by leveraging label-specific features, where a simple yet effective algorithm named LIFT is presented. Briefly, LIFT constructs features specific to each label by conducting clustering analysis on its positive and negative instances, and then performs training and testing by querying the clustering results. Extensive experiments across sixteen diversified data sets clearly validate the superiority of LIFT against other well-established multi-label learning algorithms.

In a seminal paper, Lin and Reiter introduced the notion of progression for basic action theories in the situation calculus. Earlier works by Moore, Scherl and Levesque extended the situation calculus to account for knowledge. In this paper, we study progression of knowledge in the situation calculus. We first adapt the concept of bisimulation from modal logic and extend Lin and Reiter's notion of progression to accommodate knowledge. We show that for physical actions, progression of knowledge reduces to forgetting predicates in first-order modal logic. We identify a class of first-order modal formulas for which forgetting an atom is definable in first-order modal logic. This class of formulas goes beyond formulas without quantifying-in. We also identify a simple case where forgetting a predicate reduces to forgetting a finite number of atoms. Thus we are able to show that for local-effect physical actions, when the initial KB is a formula in this class, progression of knowledge is definable in first-order modal logic. Finally, we extend our results to the multi-agent case.

To support more precise query translation for English-Chinese Bi-Directional Cross-Language Information Retrieval (CLIR), we have developed a novel framework by integrating a semantic network to characterize the correlations between multiple inter-related text terms of interest and learn their inter-related statistical query translation models. First, a semantic network is automatically generated from large-scale English-Chinese bilingual parallel corpora to characterize the correlations between a large number of text terms of interest. Second, the semantic network is exploited to learn the statistical query translation models for such text terms of interest. Finally, these inter-related query translation models are used to translate the queries more precisely and achieve more effective CLIR. Our experiments on a large number of official public data have obtained very positive results.