In an interlinked corpus of documents, the context in which a citation appears provides extra information about the cited document. However, associating terms in the context to the cited document remains an open problem. We propose a novel document generation approach that statistically incorporates the context in which a document links to another document. We quantitatively show that the proposed generation scheme explains the linking phenomenon better than previous approaches. The context information along with the actual content of the document provides significant improvements over the previous approaches for various real world evaluation tasks such as link prediction and log-likelihood estimation on unseen content. The proposed method is more scalable to large collection of documents compared to the previous approaches.

We introduce several novel word features for keyword extraction and headline generation. These new word features are derived according to the background knowledge of a document as supplied by Wikipedia. Given a document, to acquire its background knowledge from Wikipedia, we first generate a query for searching the Wikipedia corpus based on the key facts present in the document. We then use the query to find articles in the Wikipedia corpus that are closely related to the contents of the document. With the Wikipedia search result article set, we extract the inlink, outlink, category and infobox information in each article to derive a set of novel word features which reflect the document's background knowledge. These newly introduced word features offer valuable indications on individual words' importance in the input document. They serve as nice complements to the traditional word features derivable from explicit information of a document. In addition, we also introduce a word-document fitness feature to charcterize the influence of a document's genre on the keyword extraction and headline generation process. We study the effectiveness of these novel word features for keyword extraction and headline generation by experiments and have obtained very encouraging results.

Many problems require repeated inference on probabilistic graphical models, with different values for evidence variables or other changes. Examples of such problems include utility maximization, MAP inference, online and interactive inference, parameter and structure learning, and dynamic inference. Since small changes to the evidence typically only affect a small region of the network, repeatedly performing inference from scratch can be massively redundant. In this paper, we propose expanding frontier belief propagation (EFBP), an efficient approximate algorithm for probabilistic inference with incremental changes to the evidence (or model). EFBP is an extension of loopy belief propagation (BP) where each run of inference reuses results from the previous ones, instead of starting from scratch with the new evidence; messages are only propagated in regions of the network affected by the changes. We provide theoretical guarantees bounding the difference in beliefs generated by EFBP and standard BP, and apply EFBP to the problem of expected utility maximization in influence diagrams. Experiments on viral marketing and combinatorial auction problems show that EFBP can converge much faster than BP without significantly affecting the quality of the solutions.

We report a new filter for assisting the search for rare bird species. Since a rare bird only appears in front of the camera with very low occurrence (e.g. less than ten times per year) for very short duration (e.g. less than a fraction of a second), our algorithm must have very low false negative rate. We verify the bird body axis information with the known bird flying dynamics from the short video segment. Since a regular extended Kalman filter (EKF) cannot converge due to high measurement error and limited data, we develop a novel Probable Observation Data Set (PODS)-based EKF method. The new PODS-EKF searches the measurement error range for all probable observation data that ensures the convergence of the corresponding EKF in short time frame. The algorithm has been extensively tested in experiments. The results show that the algorithm achieves 95.0% area under ROC curve in physical experiment with close to zero false negative rate.

We consider settings in which voters vote in sequence, each voter knows the votes of the earlier voters and the preferences of the later voters, and voters are strategic. This can be modeled as an extensive-form game of perfect information, which we call a Stackelberg voting game. We first propose a dynamic-programming algorithm for finding the backward-induction outcome for any Stackelberg voting game when the rule is anonymous; this algorithm is efficient if the number of alternatives is no more than a constant. We show how to use compilation functions to further reduce the time and space requirements. Our main theoretical results are paradoxes for the backward-induction outcomes of Stackelberg voting games. We show that for any n ≥ 5 and any voting rule that satisfies nonimposition and with a low domination index, there exists a profile consisting of n voters, such that the backward-induction outcome is ranked somewhere in the bottom two positions in almost every voter’s preferences. Moreover, this outcome loses all but one of its pairwise elections. Furthermore, we show that many common voting rules have a very low (= 1) domination index, including all majority-consistent voting rules. For the plurality and nomination rules, we show even stronger paradoxes. Finally, using our dynamic-programming algorithm, we run simulations to compare the backward-induction outcome of the Stackelberg voting game to the winner when voters vote truthfully, for the plurality and veto rules. Surprisingly, our experimental results suggest that on average, more voters prefer the backward-induction outcome.

The problem of dividing a sequence of values into segments occurs in database systems, information retrieval, and knowledge management. The challenge is to select a finite number of boundaries for the segments so as to optimize an objective error function defined over those segments. Although this optimization problem can be solved in polynomial time, the algorithm which achieves the minimum error does not scale well, hence it is not practical for applications with massive data sets. There is considerable research with numerous approximation and heuristic algorithms. Still, none of those approaches has resolved the quality-efficiency tradeoff in a satisfactory manner. In (Halim, Karras, and Yap 2009), we obtain near linear time algorithms which achieve both the desired scalability and near-optimal quality, thus dominating earlier approaches. In this paper, we show how two ideas from artificial intelligence, an efficient local search and recombination of multiple solutions reminiscent of genetic algorithms, are combined in a novel way to obtain state of the art histogram construction algorithms.

Gaussian Mixture Model (GMM) is one of the most popular data clustering methods which can be viewed as a linear combination of different Gaussian components. In GMM, each cluster obeys Gaussian distribution and the task of clustering is to group observations into different components through estimating each cluster's own parameters. The Expectation-Maximization algorithm is always involved in such estimation problem. However, many previous studies have shown naturally occurring data may reside on or close to an underlying submanifold. In this paper, we consider the case where the probability distribution is supported on a submanifold of the ambient space. We take into account the smoothness of the conditional probability distribution along the geodesics of data manifold. That is, if two observations are close in intrinsic geometry, their distributions over different Gaussian components are similar. Simply speaking, we introduce a novel method based on manifold structure for data clustering, called Locally Consistent Gaussian Mixture Model (LCGMM). Specifically, we construct a nearest neighbor graph and adopt Kullback-Leibler Divergence as the distance measurement to regularize the objective function of GMM. Experiments on several data sets demonstrate the effectiveness of such regularization.

To function in human-inhabited environments a robot must be able to press buttons. There are literally thousands of different buttons, which produce various types of feedback when pressed. This work focuses on doorbell buttons, which provide auditory feedback. Our robot learned to predict if a specific pushing movement would press a doorbell button and produce a sound. The robot explored different buttons with random pushing behaviors and perceived the proprioceptive, tactile, and acoustic outcomes of these behaviors.

With the development of Web 2.0, sentiment analysis has now become a popular research problem to tackle. Recently, topic models have been introduced for the simultaneous analysis for topics and the sentiment in a document. These studies, which jointly model topic and sentiment, take the advantage of the relationship between topics and sentiment, and are shown to be superior to traditional sentiment analysis tools. However, most of them make the assumption that, given the parameters, the sentiments of the words in the document are all independent. In our observation, in contrast, sentiments are expressed in a coherent way. The local conjunctive words, such as “and” or “but”, are often indicative of sentiment transitions. In this paper, we propose a major departure from the previous approaches by making two linked contributions. First, we assume that the sentiments are related to the topic in the document, and put forward a joint sentiment and topic model, i.e. Sentiment-LDA. Second, we observe that sentiments are dependent on local context. Thus, we further extend the Sentiment-LDA model to Dependency-Sentiment-LDA model by relaxing the sentiment independent assumption in Sentiment-LDA. The sentiments of words are viewed as a Markov chain in Dependency-Sentiment-LDA. Through experiments, we show that exploiting the sentiment dependency is clearly advantageous, and that the Dependency-Sentiment-LDA is an effective approach for sentiment analysis.

The rectangle packing problem consists of finding an enclosing rectangle of smallest area that can contain a given set of rectangles without overlap. We propose two new benchmarks, one where the orientation of the rectangles is fixed and one where it is free, that include rectangles of various aspect ratios. The new benchmarks avoid certain properties of easy instances, which we identify as instances where rectangles have dimensions in common or where a few rectangles occupy most of the area. Our benchmarks are much more difficult for the previous state-of-the-art solver, requiring orders of magnitude more time, compared to similar-sized instances from a popular benchmark consisting only of squares. On the new benchmarks, we improve upon the previous strategy used to handle dominance conditions, we define a variable order over non-square rectangles that generalizes previous strategies, and we present a way to adjust the sizes of intervals of values for each rectangle's x-coordinates. Using these techniques together, we can solve the new oriented benchmark about 500 times faster, and the new unoriented benchmark about 40 times faster than the previous state-of-the-art.