Learning to Perceive and Act by Trial and Error

This article considers adaptive control architectures that integrate active sensory-motor systems with decision systems based on reinforcement learning. One unavoidable consequence of active perception is that the agent's internal representation often confounds external world states. We call this phoenomenon perceptual aliasingand show that it destabilizes existing reinforcement learning algorithms with respect to the optimal decision policy. We then describe a new decision system that overcomes these difficulties for a restricted class of decision problems. The system incorporates a perceptual subcycle within the overall decision cycle and uses a modified learning algorithm to suppress the effects of perceptual aliasing. The result is a control architecture that learns not only how to solve a task but also where to focus its visual attention in order to collect necessary sensory information.     

On the learning control of a robot manipulator

This paper derives a learning control law to achieve trajectory following for a robot manipulator. The controller consists of two parts, a computed torque servo for the rigid body terms that can be modelled and a learning law for the unmodelled dynamics. An advantage of this method is that bounds can be assigned to the position and velocity tracking errors.     

Learning to Predict Non-Deterministically Generated Strings

In this article we present an algorithm that learns to predict non-deterministically generated strings. The problem of learning to predict non-deterministically generated strings was raised by Dietterich and Michalski (1986). While their objective was to give heuristic techniques that could be used to rapidly and effectively learn to predict a somewhat limited class of strings, our objective is to give an algorithm which, though impractical, is capable of learning to predict a very general class. Our algorithm is meant to provide a general framework within which heuristic techniques can be effectively employed.     

Extending boosting for large scale spoken language understanding

We propose three methods for extending the Boosting family of classifiers motivated by the real-life problems we have encountered. First, we propose a semisupervised learning method for exploiting the unlabeled data in Boosting. We then present a novel classification model adaptation method. The goal of adaptation is optimizing an existing model for a new target application, which is similar to the previous one but may have different classes or class distributions. Finally, we present an efficient and effective cost-sensitive classification method that extends Boosting to allow for weighted classes. We evaluated these methods for call classification in the AT&;T VoiceTone^® spoken language understanding system. Our results indicate that it is possible to obtain the same classification performance by using 30% less labeled data when the unlabeled data is utilized through semisupervised learning. Using model adaptation we can achieve the same classification accuracy using less than half of the labeled data from the new application. Finally, we present significant improvements in the “important” (i.e., higher weighted) classes without a significant loss in overall performance using the proposed cost-sensitive classification method.     

Age of acquisition: Its neural and computational mechanisms.

The acquisition of new skills over a life span is a remarkable human ability. This ability, however, is constrained by age of acquisition (AoA); that is, the age at which learning occurs significantly affects the outcome. This is most clearly reflected in domains such as language, music, and athletics. This article provides a perspective on the neural and computational mechanisms underlying AoA in language acquisition. The authors show how AoA modulates both monolingual lexical processing and bilingual language acquisition. They consider the conditions under which syntactic processing and semantic processing may be differentially sensitive to AoA effects in second-language acquisition. The authors conclude that AoA effects are pervasive and that the neural and computational mechanisms underlying learning and sensorimotor integration provide a general account of these effects. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

David W. Johnson: Award for Distinguished Contributions of Applications of Psychology to Education and Training.

Provides the biography of David W. Johnson and announces that he has won the Award for Distinguished Contributions of Applications of Psychology to Education and Training for programmatic research, comprehensive research reviews, and revision of theory over four decades of persistence that have provided the knowledge foundation for cooperative learning, constructive controversy, conflict resolution, and peer mediation training. A selected bibliography is also provided. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Revising threshold functions

A revision algorithm is a learning algorithm that identifies the target concept, starting from an initial concept. Such an algorithm is considered efficient if its complexity (in terms of the resource one is interested in) is polynomial in the syntactic distance between the initial and the target concept, but only polylogarithmic in the number of variables in the universe. We give an efficient revision algorithm in the model of learning with equivalence and membership queries for threshold functions, and some negative results showing, for instance, that threshold functions cannot be revised efficiently from either type of query alone. The algorithms work in a general revision model where both deletion and addition type revision operators are allowed.     

Interactive simulated patient—an advanced tool for student-activated learning in medicine and healthcare

An advanced learning system for interactive simulation of patient cases (ISP) is described. The system was designed to meet specific pedagogical needs where student activation and problem orientation were two of the key issues. The system includes features such as video based illness history-taking function using free text input, highly interactive physiological examination procedures, extensive laboratory tests and detailed user feedback. The system was constructed mainly in Authorware Professional, starting more than 10 years ago, and it is now available in refined versions in different languages. Twelve pedagogical and three technical features were initially defined for the project. All of these aims were fulfilled. The ISP system has been successfully used in education, and in international collaboration, including three applicability field tests. The system is also a powerful research tool for studying for instance medical decision-making.     

Context-gated knowledge partitioning in categorization.

According to the knowledge partitioning framework, people sometimes master complex tasks by creating multiple independent parcels of partial knowledge. Research has shown that knowledge parcels may contain mutually contradictory information, and that each parcel may be used without regard to knowledge that is demonstrably present in other parcels. This article reports 4 experiments that investigated knowledge partitioning in categorization. When component boundaries of a complex categorization were identified by a context cue, a significant proportion of participants learned partial and independent categorization strategies that were chosen on the basis of context. For those participants, a strategy used in one context was unaffected by knowledge demonstrably present in other contexts, suggesting that knowledge partitioning in categorization can be complete. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

神经网络系统学习过程初探

本文给出了神经网络系统学习过程的一个随机模型--马尔柯夫模型,从最大熵原理的 观点讨论神经网络系统的学习过程,提出将神经网络系统的学习过程分为两个阶段的想法并 给出学习过程的算法.