We present a system that enables rapid model experimentation for tera-scale machine learning with trillions of non-zero features, billions of training examples, and millions of parameters. Our contribution to the literature is a new method (SA L-BFGS) for changing batch L-BFGS to perform in near real-time by using statistical tools to balance the contributions of previous weights, old training examples, and new training examples to achieve fast convergence with few iterations. The result is, to our knowledge, the most scalable and flexible linear learning system reported in the literature, beating standard practice with the current best system (Vowpal Wabbit and AllReduce). Using the KDD Cup 2012 data set from Tencent, Inc. we provide experimental results to verify the performance of this method.

We introduce a class of learning problems where the agent is presented with a series of tasks. Intuitively, if there is relation among those tasks, then the information gained during execution of one task has value for the execution of another task. Consequently, the agent is intrinsically motivated to explore its environment beyond the degree necessary to solve the current task it has at hand. We develop a decision theoretic setting that generalises standard reinforcement learning tasks and captures this intuition. More precisely, we consider a multi-stage stochastic game between a learning agent and an opponent. We posit that the setting is a good model for the problem of life-long learning in uncertain environments, where while resources must be spent learning about currently important tasks, there is also the need to allocate effort towards learning about aspects of the world which are not relevant at the moment. This is due to the fact that unpredictable future events may lead to a change of priorities for the decision maker. Thus, in some sense, the model "explains" the necessity of curiosity. Apart from introducing the general formalism, the paper provides algorithms. These are evaluated experimentally in some exemplary domains. In addition, performance bounds are proven for some cases of this problem.

We present a new analysis of the problem of learning with drifting distributions in the batch setting using the notion of discrepancy. We prove learning bounds based on the Rademacher complexity of the hypothesis set and the discrepancy of distributions both for a drifting PAC scenario and a tracking scenario. Our bounds are always tighter and in some cases substantially improve upon previous ones based on the $L_1$ distance. We also present a generalization of the standard on-line to batch conversion to the drifting scenario in terms of the discrepancy and arbitrary convex combinations of hypotheses. We introduce a new algorithm exploiting these learning guarantees, which we show can be formulated as a simple QP. Finally, we report the results of preliminary experiments demonstrating the benefits of this algorithm.

A vast amount of textual web streams is influenced by events or phenomena emerging in the real world. The social web forms an excellent modern paradigm, where unstructured user generated content is published on a regular basis and in most occasions is freely distributed. The present Ph.D. Thesis deals with the problem of inferring information - or patterns in general - about events emerging in real life based on the contents of this textual stream. We show that it is possible to extract valuable information about social phenomena, such as an epidemic or even rainfall rates, by automatic analysis of the content published in Social Media, and in particular Twitter, using Statistical Machine Learning methods. An important intermediate task regards the formation and identification of features which characterise a target event; we select and use those textual features in several linear, non-linear and hybrid inference approaches achieving a significantly good performance in terms of the applied loss function. By examining further this rich data set, we also propose methods for extracting various types of mood signals revealing how affective norms - at least within the social web's population - evolve during the day and how significant events emerging in the real world are influencing them. Lastly, we present some preliminary findings showing several spatiotemporal characteristics of this textual information as well as the potential of using it to tackle tasks such as the prediction of voting intentions.

ARTINGALES are one of the fundamental tools in probability theory and statistics for modeling and studying sequences of random variables. Some of the most well-known and widely used concentration inequalities for individual martingales are Hoeffding-Azuma's and Bernstein's inequalities [1], [2], [3]. We present a comparison inequality that bounds the expectation of a convex function of a martingale difference sequence shifted to the [0, 1] interval by the expectation of the same function of independent Bernoulli variables. We apply this inequality in order to derive a tighter analog of Hoeffding-Azuma's inequality for martingales. More importantly, we present a set of inequalities that make it possible to control weighted averages of multiple simultaneously evolving and interdependent martingales (see Figure 1 for an illustration).

Our long-term goal is to develop autonomous robotic systems that have the cognitive abilities of humans, including communication, coordination, adapting to novel situations, and learning through experience. Our approach rests on the integration of the Soar cognitive architecture with both virtual and physical robotic systems. Soar has been used to develop a wide variety of knowledge-rich agents for complex virtual environments, including distributed training environments and interactive computer games. For development and testing in robotic virtual environments, Soar interfaces to a variety of robotic simulators and a simple mobile robot. We have recently made significant extensions to Soar that add new memories and new non-symbolic reasoning to Soar’s original symbolic processing, which improves Soar abilities for control of robots. These extensions include mental imagery, episodic and semantic memory, reinforcement learning, and continuous model learning. This paper presents research in mobile robotics, relational and continuous model learning, and learning by situated, interactive instruction.

Interest in online education is surging, as dramatized bythe success of Khan Academy and recent Stanford online courses, but the technology for online education isin its infancy. Crowdsourcing mechanisms will likelybe essential in order to reach the full potential of thismedium. This paper sketches some of the challengesand directions we hope HCOMP researchers will address.

This paper describes the development and empirical testing of an intelligent tutoring system (ITS) with two emerging methodologies: (1) a partially observable Markov decision process (POMDP) for representing the learner model and (2) inquiry modeling, which informs the learner model with questions learners ask during instruction. POMDPs have been successfully applied to non-ITS domains but, until recently, have seemed intractable for large-scale intelligent tutoring challenges. New, ITS-specific representations leverage common regularities in intelligent tutoring to make a POMDP practical as a learner model. Inquiry modeling is a novel paradigm for informing learner models by observing rich features of learners’ help requests such as categorical content, context, and timing. The experiment described in this paper demonstrates that inquiry modeling and planning with POMDPs can yield significant and substantive learning improvements in a realistic, scenario-based training task.

Qualitative representations are suitable for sketch understanding systems because they highlight important relationships while leaving out details that are not essential for conceptual understanding. These representations can be used to perform spatial analogies between sketches, which determine qualitative similarities and differences. However, there are cases where including quantitative information is necessary for accurately representing a sketch. We describe a method for using quantitative information to constrain qualitative spatial analogies. The utility of this method is demonstrated in the context of a sketch-based educational software system. Importantly, using quantitative information to improve analogical matches is not domain-specific. It can be used in any situation where qualitative and quantitative spatial information must be combined to accurately interpret a sketch. This approach has the potential to improve sketch understanding in educational software applications for highly spatial domains.

Introductory engineering courses within large universities often have annual enrollments which can reach up to a thousand students. It is very challenging to achieve differentiated instruction in classrooms with class sizes and student diversity of such great magnitude. Professors can only assess whether students have mastered a concept by using multiple choice questions, while detailed homework assignments, such as planar truss diagrams, are rarely assigned because professors and teaching assistants would be too overburdened with grading to return assignments with valuable feedback in a timely manner. In this paper, we introduce Mechanix, a sketch-based deployed tutoring system for engineering students enrolled in statics courses. Our system not only allows students to enter planar truss and free body diagrams into the system just as they would with pencil and paper, but our system checks the student's work against a hand-drawn answer entered by the instructor, and then returns immediate and detailed feedback to the student. Students are allowed to correct any errors in their work and resubmit until the entire content is correct and thus all of the objectives are learned. Since Mechanix facilitates the grading and feedback processes, instructors are now able to assign free response questions, increasing teacher's knowledge of student comprehension. Furthermore, the iterative correction process allows students to learn during a test, rather than simply displaying memorized information.