For nearly 25 years my research group has investigated the use of domain knowledge, expressed in some version of mathematical logic, that is refined or exploited by numeric-based learning algorithms. These include what we called knowledge-based neural networks and knowledge-based support vector machines. I will cover the key ideas of these methods, as well as the behind-the-scenes motivations that lead to them. I will also describe why we switched from using the phrase 'prior knowledge' to using 'advice.' Finally, I will cover some of our recent work on fast learning and inference for Markov Logic Networks (which can be viewed as a knowledge-based graphical model).

We present a new weighted session similarity measure to capture the browsing interests of users in web usage profiles discovered from web log data. We base our similarity measure on the reasonable assumption that when users spend longer times on pages or revisit pages in the same session, then very likely, such pages are of greater interest to the user. The proposed similarity measure combines structural similarity with session-wise page significance. The latter, representing the degree of user interest, is computed using frequency and duration of a page access. Web usage profiles are generated using this similarity measure by applying a fuzzy clustering algorithm to web log data. For evaluating the effectiveness of the proposed measure, we adapt two model-based collaborative filtering algorithms for recommending pages. Experimental results show considerable improvement in overall performance of recommender systems as compared to use of other existing similarity measures.

Research over the past years has shown that machine translation results can be greatly enhanced with the help of mono- or bilingual human contributors, e.g. by asking hu- mans to proofread or correct outputs of machine translation systems. However, it remains difficult to determine the quality of individual revisions. This paper proposes a meth- od to determine the quality of individual contributions by analyzing task-independent data. Examples of such data are completion time, number of keystrokes, etc. An initial evaluation showed promising F-measure values larger than 0.8 for support vector machine and decision tree based classifications of a combined test set of Vietnamese and German translations.

Computational systems can use sociocultural knowledge to understand human behavior and interact with humans in more natural ways. However, such systems are limited by their reliance on hand-authored sociocultural knowledge and models. We introduce an approach to automatically learn robust, script-like sociocultural knowledge from crowdsourced narratives. Crowdsourcing, the use of anonymous human workers, provides an opportunity for rapidly acquir­ing a corpus of examples of situations that are highly specialized for our purpose yet sufficiently varied, from which we can learn a versatile script. We describe a semi-automated process by which we query human workers to write natural language narrative examples of a given situation and learn the set of events that can occur and the typical even ordering.

Crowdsourcing services are often used to collect a large amount of labeled data for machine learning. Although they provide us an easy way to get labels at very low cost in a short period, they have serious limitations. One of them is the variable quality of the crowd-generated data. There have been many attempts to increase the reliability of crowd-generated data and the quality of classifiers obtained from such data. However, in these problem settings, relatively few researchers have tried using expert-generated data to achieve further improvements. In this paper, we extend three models that deal with the problem of learning from crowds to utilize ground truths: a latent class model, a personal classifier model, and a data-dependent error model. We evaluate the proposed methods against two baseline methods on a real data set to demonstrate the effectiveness of combining crowd-generated data and expert-generated data.

Crowdsourcing utilizes human ability by distributing tasks to a large number of workers. It is especially suitable for solving data clustering problems because it provides a way to obtain a similarity measure between objects based on manual annotations, which capture the human perception of similarity among objects.This is in contrast to most clustering algorithms that face the challenge of finding an appropriate similarity measure for the given dataset. Several algorithms have been developed for crowdclustering that combine partial clustering results, each obtained by annotations provided by a different worker, into a single data partition. However, existing crowd-clustering approaches require a large number of annotations, due to the noisy nature of human annotations, leading to a high computational cost in addition to the large cost associated with annotation. We address this problem by developing a novel approach for crowclustering that exploits the technique of matrix completion. Instead of using all the annotations, the proposed algorithm constructs a partially observed similarity matrix based on a subset of pairwise annotation labels that are agreed upon by most annotators. It then deploys the matrix completion algorithm to complete the similarity matrix and obtains the final data partition by applying a spectral clustering algorithm to the completed similarity matrix. We show, both theoretically and empirically, that the proposed approach needs only a small number of manual annotations to obtain an accurate data partition. In effect, we highlight the trade-off between a large number of noisy crowdsourced labels and a small number of high quality labels.

We propose an approach for the representation of actions based on the conceptual spaces framework developed by Gärdenfors (2004). Action categories are regarded as properties in the sense of Gärdenfors (2011) and are understood as convex regions in action space. Action categories are mainly described by a force signature that represents the forces that act upon a main trajector involved in the action. This force signature is approximated via a representation that specifies the time-indexed position of the trajector relative to several landmarks. We also present a computational approach to extract such representations from video data. We present results on the Motionese dataset consisting of videos of parents demonstrating actions on objects to their children. We evaluate the representations on a clustering and a classification task showing that, while our representations seems to be reasonable, only a handful of actions can be discriminated reliably.

In digital libraries, a table, as a specific document component as well as a condensed way to present structured and relational data, contains rich information and often the only source of .that information. In order to explore, retrieve, and reuse that data, tables should be identified and the data extracted. Table recognition is an old field of research. However, due to the diversity of table styles, the results are still far from satisfactory, and not a single algorithm performs well on all different types of tables. In this paper, we randomly take samples from the CiteSeerX to investigate diverse table styles for automatic table extraction. We find that table headers are one of the main characteristics of complex table styles. We identify a set of features that can be used to segregate headers from tabular data and build a classifier to detect table headers. Our empirical evaluation on PDF documents shows that using a Random Forest classifier achieves an accuracy of 92%.

These results far outperform alternative models. This work is an important step towards the development Recent work has demonstrated that micro-blogging data can of automated methods that identify disease vectors, trace the be used to predict a variety of phenomena, including movie transmission between concrete individuals, and ultimately box-office revenues (Asur and Huberman 2010), elections help us understand and predict the spread of infectious diseases (Tumasjan et al. 2010), and flu epidemics (Lampos, De Bie, with fine granularity. It provides a foundation for and Cristianini 2010). Most research to date has focused on research on fundamental questions of public health, such predicting aggregate properties of the population from the as: How does an epidemic on a population scale emerge activity of the bloggers. A different kind of problem one can from low-level interactions between people in the course pose, however, is to predict the behavior or state of particular of their everyday lives? Can we identify a potentially noncooperative individuals within the social network. For instance, one individual who is a vector of a dangerous disease, could try to predict whether a person will go to a movie or i.e., a "Typhoid Mary"? What is the interaction between vote for a particular candidate based on micro-blog data. The friendship, location, and co-location in the spread of individual's own data may or may not be accessible.

The step-size, often denoted as α, is a key parameter for most incremental learning algorithms. Its importance is especially pronounced when performing online temporal difference (TD) learning with function approximation. Several methods have been developed to adapt the step-size online. These range from straightforward back-off strategies to adaptive algorithms based on gradient descent. We derive an adaptive upper bound on the step-size parameter to guarantee that online TD learning with linear function approximation will not diverge. We then empirically evaluate algorithms using this upper bound as a heuristic for adapting the step-size parameter online. We compare performance with related work including HL(λ) and Autostep. Our results show that this adaptive upper bound heuristic out-performs all existing methods without requiring any meta-parameters. This effectively eliminates the need to tune the learning rate of temporal difference learning with linear function approximation.