Semantic role labeling (SRL) is a technique of semantic interpretation of text on the sentence level. In this paper, we present a corpus that is labeled with semantic roles for biological transport events. The corpus was built using domain knowledge provided by ontologies. We also report on a word-chunking approach for identifying semantic roles of biomedical predicates describing transport events. We trained a first-order Conditional Random Fields (CRF) for chunking applications with the traditional role labeling features and also domain-specific features. The results show that the system performance varies between different roles and the performance was not improved for all roles by introducing domain specific features.

Learning from demonstration (LfD) is a promising technique for instructing/teaching autonomous systems based on demonstrations from people who may have little to no experience with robots. An important aspect to LfD is the communication method used to transfer knowledge from an instructor to a robot. The communication method affects the complexity of the demonstration process for instructors, the range of tasks a robot can learn, and the learning algorithm itself. We have designed a graphical interface and an instructional language to provide an intuitive teaching system. The drawback to simplifying the teaching interface is that the resulting demonstration data are less structured, adding complexity to the learning process. This additional complexity is handled through the combination of a minimal set of predefined behaviors and a task representation capable of learning probabilistic policies over a set of behaviors. The predefined behaviors consist of finite actions a robot can perform, which act as building blocks for more complex tasks.

Human interaction with robot swarms (HSI) is a young field with very few user studies that explore operator behavior. All these studies assume perfect communication between the operator and the swarm. A key challenge in the use of swarm robotic systems in human supervised tasks is to understand human swarm interaction in the presence of limited communication bandwidth, which is a constraint arising in many practical scenarios. In this paper, we present results of human-subject experiments designed to study the effect of bandwidth limitations in human swarm interaction. We consider three levels of bandwidth availability in a swarm foraging task. The lowest bandwidth condition performs poorly, but the medium and high bandwidth condition both perform well. In the medium bandwidth condition, we display useful aggregated swarm information (like swarm centroid and spread) to compress the swarm state information. We also observe interesting operator behavior and adaptation of operators’ swarm reaction.

Many methods have been proposed for making use of multiple experts to predict uncertain events such as election outcomes, ranging from simple averaging of individual predictions to complex collaborative structures such as prediction markets or structured group decision making processes. We used a panel of more than 2,000 forecasters to systematically compare the performance of four different collaborative processes on a battery of political prediction problems. We found that teams and prediction markets systematically outperformed averages of individual forecasters, that training forecasters helps, and that the exact form of how predictions are combined has a large effect on overall prediction accuracy.

Robots are increasingly introduced to work in concert with people in high-intensity domains, such as manufacturing, space exploration and hazardous environments. Although there are numerous studies on human teamwork and coordination in these settings, very little prior work exists on applying these models to human-robot interaction. This paper presents results from ongoing work aimed at translating qualitative methods from human factors engineering into computational models that can be applied to human-robot teaming. We describe a statistical approach to learning patterns of strong and weak agreements in human planning meetings that achieves up to 94% prediction accuracy. We also formulate a human-robot interactive planning method that emulates cross-training, a training strategy widely used in human teams. Results from human subject experiments show statistically significant improvements on team fluency metrics, compared to standard reinforcement learning techniques. Results from these two studies support the approach of modeling and applying common practices in human teaming to achieve more effective and fluent human-robot teaming.

Clinical research of new recreational drugs and trends requires mining current information from non-traditional text sources. In this work we support such research through the use of multi-dimensional latent text models, such as factorial LDA, that capture orthogonal factors of corpora, creating structured output for researchers to better understand the contents of a corpus. Since a purely unsupervised model is unlikely to discover specific factors of interests to clinical researchers, we modify the structure of factorial LDA to incorporate prior knowledge, including the use of of observed variables, informative priors and background components. The resulting model learns factors that correspond to drug type, delivery method (smoking, injection, etc.), and aspect (chemistry, culture, effects, health, usage). We demonstrate that the improved model yields better quantitative and more interpretable results.

This talk will analyze responses to funniness of five versions of 10 different jokes. The responses of one of them will then be compared to theoretical analysis and representation of the same joke based on Script-based Semantics Theory of Humor, General Theory of Verbal Humor, and Ontological Semantic Theory of Humor.

Causality discovery has been one of the core tasks in scientific research since the beginning of human scientific history. In the age of data tsunami, the causality discovery task involves identification of causality among millions of variables which cannot be done manually by humans. However, the identification of causality relationships using artificial intelligence and statistical techniques in non-experimental settings faces several challenges. In this work, we address three of the challenges regarding Granger causality, one of the most popular causality inference techniques. First, we analyze the consistency of two most popular Granger causality techniques and show that the significance test is not consistent in high dimensions. Second, we review our nonparametric generalization of the Lasso-Granger technique called Generalized Lasso Granger (GLG) to uncover Granger causality relationships among irregularly sampled time series. Finally, we describe two techniques to uncover the casual dependence in non-linear datasets. Extensive experiments are provided to show the significant advantages of the proposed algorithms over their state-of-the-art counterparts.

In English, “puns” are usually perceived as a subclass of “jokes”. In Japanese, however, this is not necessarily true. In this paper we investigate whether Japanese native speakers perceive dajare (puns) as jooku (jokes). We first summarize existing research in the field of computational humor, both in English and Japanese, focusing on the usage of these two terms. This shows that in works of Japanese native speakers, puns are not commonly treated as jokes. Next we present some dictionary definitions of dajare and jooku, which show that they may actually be used in a similar manner to English. In order to study this issue, we conducted a survey, in which we asked Japanese participants three questions: whether they like jokes (jooku), whether they like puns (dajare) and whether dajare are jooku. The results showed that there is no common agreement regarding dajare being a genre of jokes. We analyze the outcome of this experiment and discuss them from different points of view.

There is strong theoretical evidence that aggregation of human judgments should not simply average multiple forecasts together (the unweighted linear opinion pool, or ULinOP), but weight them in such a way as to insure representation of a maximally diverse set of models among the experts from whom they are elicited. Explicitly eliciting these models places a major burden on the experts. We report on a variety of approaches to estimating these models, or at least the diversity among them, with minimal explicit input from the experts.