Better than being there: next millennium networks

At the culmination of the 20th Century, people in universities and research laboratories have achieved networks of sufficient bandwidth to effectively eliminate the need to physically ship documents, images, and videotapes. Will the next millennium bring us networks that deliver services so efficiently and expediently that we can claim they are, for us, better than being there? As professionals in computer graphics and applications, it is critical that we obtain the quality of image, sound, and interaction over distance that we can enjoy locally     

Understanding contexts by being there: case studies in bodystorming

A thorough appreciation of physical, social, interactional, and psychological contextual factors is crucial in the design of ubiquitous computing applications. This paper investigates the benefits of a method called bodystorming for carrying out design sessions in the original context, `in the wild', instead of the office. A location is selected that is identical or similar to the original environment. Innovation, carried out on-site, is based on ethnographical data presented as concrete design questions. Individual solutions to design questions are brainstormed and discussed on-site. Facets of data collection and preparation, formulation of design questions, selection of locations, session administration, and evaluation of design ideas are presented. We found that bodystorming permits immediate feedback for generated design ideas, and can provide a more accurate understanding of contextual factors. Bodystorming sessions were found memorable and inspiring. It is best suitable for designing for activities that are accessible and unfamiliar to the researchers.     

Distance learning for similarity estimation

In this paper, we present a general guideline to find a better distance measure for similarity estimation based on statistical analysis of distribution models and distance functions. A new set of distance measures are derived from the harmonic distance, the geometric distance, and their generalized variants according to the Maximum Likelihood theory. These measures can provide a more accurate feature model than the classical Euclidean and Manhattan distances. We also find that the feature elements are often from heterogeneous sources that may have different influence on similarity estimation. Therefore, the assumption of single isotropic distribution model is often inappropriate. To alleviate this problem, we use a boosted distance measure framework that finds multiple distance measures which fit the distribution of selected feature elements best for accurate similarity estimation. The new distance measures for similarity estimation are tested on two applications: stereo matching and motion tracking in video sequences. The performance of boosted distance measure is further evaluated on several benchmark data sets from the UCI repository and two image retrieval applications. In all the experiments, robust results are obtained based on the proposed methods.     

Distance metric learning by knowledge embedding

This paper presents an algorithm which learns a distance metric from a data set by knowledge embedding and uses the new distance metric to solve nonlinear pattern recognition problems such a clustering.     

Distance metric learning by minimal distance maximization

Classic linear dimensionality reduction (LDR) methods, such as principal component analysis (PCA) and linear discriminant analysis (LDA), are known not to be robust against outliers. Following a systematic analysis of the multi-class LDR problem in a unified framework, we propose a new algorithm, called minimal distance maximization (MDM), to address the non-robustness issue. The principle behind MDM is to maximize the minimal between-class distance in the output space. MDM is formulated as a semi-definite program (SDP), and its dual problem reveals a close connection to “weighted” LDR methods. A soft version of MDM, in which LDA is subsumed as a special case, is also developed to deal with overlapping centroids. Finally, we drop the homoscedastic Gaussian assumption made in MDM by extending it in a non-parametric way, along with a gradient-based convex approximation algorithm to significantly reduce the complexity of the original SDP. The effectiveness of our proposed methods are validated on two UCI datasets and two face datasets.     

Distance metric learning for graph structured data

Machine Learning - Graphs are versatile tools for representing structured data. As a result, a variety of machine learning methods have been studied for graph data analysis. Although many such...     

Distance learning immersive environments: Sense of presence exploration

Distance learning technologies are becoming more important for the delivery of content as distance education seems rapidly growing adoption worldwide. These technologies allow collaboration between students and teachers in globally disparate environments. Immersive environments are one recent innovation with significant promise, offering a sense of presence—the sense of “being there”—which can strengthen and reinforce distance learning. The primary purpose of this study is to research the sense of presence using an immersive environment. While this is a work-in-progress, the authors provide preliminary results and conclusions.     

Distance learning: From multiple snapshots, a composite portrait

This article discusses the current state of distance learning in composition by reporting on and interpreting a 2005 survey that assesses trends and workload conditions in distance learning. Areas examined in the article include attitudes of faculty and administration, faculty demographics, student demographics, online course and program development, course caps, course delivery and management tools, technology support, course design freedom, impact on writing pedagogy, and institutional DE profile. The article concludes by summarizing the current DL picture, identifying areas of need, and providing research recommendations for the future.     

Single sourcing versus dual sourcing under conditions of learning

This paper considers a buyer who has to decide whether to select a single or two sources of supply for a homogeneous product. The production processes of the suppliers are subject to learning effects, which reduce the production costs and increase the production capacities of the suppliers. This, in turn, enables the suppliers to reduce the sales price, which results in lower acquisition costs at the buyer. As the supplier selection decision influences the individual production quantity of a supplier, the learning effect has to be considered when deciding how many and which suppliers to select. Since the effect of learning on the supplier selection problem has not been investigated in the literature, this paper addresses this limitation and derives models for continuous learning and when learning plateaus. Numerical results indicate that the supplier selection decision can comprehensively influence the learning process for the suppliers and therewith the total costs of the system under study. The results also show that it is not necessarily optimal solely to select the supplier with the highest learning rate.     

Students' experience of component versus integrated virtual learning environments

Abstract The use of virtual learning environments (VLEs) has become increasingly common in most higher education (HE) institutions. Recent developments have proposed the interoperability of software systems and content, to create component VLEs in contrast with the integrated, monolithic ones that are currently prevalent. This paper examines the student experience of two VLEs, one integrated approach and the other component. In general, students preferred the component system, although this may have been influenced by other factors such as performance. Although the study is limited to one cohort of student it makes a number of suggestions relevant to anyone deploying a VLE. These are that the component approach is a viable one from a student perspective, the broader context in which the VLE operates is important in student perception and that poor system performance may have unpredictable consequences for the learning experience.     

On being incoherent without being very hard

Trakhtenbrot calls a setA autoreducible ifA is Turing-reducible toA via an oracle Turing machine that never queries the oracle about the input string. Yao considers sets that are autoreducible via probabilistic, polynomial-time oracle Turing machines, and he calls such setscoherent. We continue the study of autoreducible sets, including those that are autoreducible via a family of polynomial-sized circuits, which we callweakly coherent. Sets that are not weakly coherent are calledstrongly incoherent. We show

Distance metric learning-based kernel gram matrix learning for pattern analysis tasks in kernel feature space

Approaches to distance metric learning (DML) for Mahalanobis distance metric involve estimating a parametric matrix that is associated with a linear transformation. For complex pattern analysis tasks, it is necessary to consider the approaches to DML that involve estimating a parametric matrix that is associated with a nonlinear transformation. One such approach involves performing the DML of Mahalanobis distance in the feature space of a Mercer kernel. In this approach, the problem of estimation of a parametric matrix of Mahalanobis distance is formulated as a problem of learning an optimal kernel gram matrix from the kernel gram matrix of a base kernel by minimizing the logdet divergence between the kernel gram matrices. We propose to use the optimal kernel gram matrices learnt from the kernel gram matrix of the base kernels in pattern analysis tasks such as clustering, multi-class pattern classification and nonlinear principal component analysis. We consider the commonly used kernels such as linear kernel, polynomial kernel, radial basis function kernel and exponential kernel as well as hyper-ellipsoidal kernels as the base kernels for optimal kernel learning. We study the performance of the DML-based class-specific kernels for multi-class pattern classification using support vector machines. Results of our experimental studies on benchmark datasets demonstrate the effectiveness of the DML-based kernels for different pattern analysis tasks.     

Online calibrated forecasts: Memory efficiency versus universality for learning in games

We provide a simple learning process that enables an agent to forecast a sequence of outcomes. Our forecasting scheme, termed tracking forecast, is based on tracking the past observations while emphasizing recent outcomes. As opposed to other forecasting schemes, we sacrifice universality in favor of a significantly reduced memory requirements. We show that if the sequence of outcomes has certain properties—it has some internal (hidden) state that does not change too rapidly—then the tracking forecast is weakly calibrated so that the forecast appears to be correct most of the time. For binary outcomes, this result holds without any internal state assumptions. We consider learning in a repeated strategic game where each player attempts to compute some forecast of the opponent actions and play a best response to it. We show that if one of the players uses a tracking forecast, while the other player uses a standard learning algorithm (such as exponential regret matching or smooth fictitious play), then the player using the tracking forecast obtains the best response to the actual play of the other players. We further show that if both players use tracking forecast, then under certain conditions on the game matrix, convergence to a Nash equilibrium is possible with positive probability for a larger class of games than the class of games for which smooth fictitious play converges to a Nash equilibrium.     

Distance learning in advanced military education: Analysis of joint operations course in the Taiwan military

High-ranking officers require advanced military education in war tactics for future combat. However, line officers rarely have time to take such courses on campus. The conventional solution to this problem used to take the inefficient correspondence courses. Whereas Internet technologies progress, online course is the current trend for military training. However, the question is what distance learning methodology best suits such a proprietary learning purpose.This study presents a sequential process of developing distance learning courses in advanced military education. Further, the Petri-Net analytical approach is adopted to discover the essential interaction requirements of advanced military education delivered via Internet. This study developed a systematic method for designing e-learning systems according to specific requirements of target courses. The proposed approach starts by comparing on-campus programs with the existing e-learning systems to identify the steps required to transform the program into an e-learning system. After first outlining the pedagogy of the on-campus program, its proposed teaching flow through the Internet is then sketched. Finally, the Petri-Net model was used for in-depth analysis of the stages affecting the learning curve of the line office taking e-learning courses. The example of a “Joint Operations” AME course elaborated the presented approach. An e-learning system prototype was also designed accordingly. Lastly, an experiment was conducted to verify the efficiency of the presented approach.     

Coevolution versus self-play temporal difference learning for acquiring position evaluation in small-board go

Two learning methods for acquiring position evaluation for small Go boards are studied and compared. In each case the function to be learned is a position-weighted piece counter and only the learning method differs. The methods studied are temporal difference learning (TDL) using the self-play gradient-descent method and coevolutionary learning, using an evolution strategy. The two approaches are compared with the hope of gaining a greater insight into the problem of searching for "optimal" zero-sum game strategies. Using tuned standard setups for each algorithm, it was found that the temporal-difference method learned faster, and in most cases also achieved a higher level of play than coevolution, providing that the gradient descent step size was chosen suitably. The performance of the coevolution method was found to be sensitive to the design of the evolutionary algorithm in several respects. Given the right configuration, however, coevolution achieved a higher level of play than TDL. Self-play results in optimal play against a copy of itself. A self-play player will prefer moves from which it is unlikely to lose even when it occasionally makes random exploratory moves. An evolutionary player forced to perform exploratory moves in the same way can achieve superior strategies to those acquired through self-play alone. The reason for this is that the evolutionary player is exposed to more varied game-play, because it plays against a diverse population of players.     

The actor-critic learning is behind the matching law: matching versus optimal behaviors

The ability to make a correct choice of behavior from various options is crucial for animals' survival. The neural basis for the choice of behavior has been attracting growing attention in research on biological and artificial neural systems. Alternative choice tasks with variable ratio (VR) and variable interval (VI) schedules of reinforcement have often been employed in studying decision making by animals and humans. In the VR schedule task, alternative choices are reinforced with different probabilities, and subjects learn to select the behavioral response rewarded more frequently. In the VI schedule task, alternative choices are reinforced at different average intervals independent of the choice frequencies, and the choice behavior follows the so-called matching law. The two policies appear robustly in subjects' choice of behavior, but the underlying neural mechanisms remain unknown. Here, we show that these seemingly different policies can appear from a common computational algorithm known as actor-critic learning. We present experimentally testable variations of the VI schedule in which the matching behavior gives only a suboptimal solution to decision making and show that the actor-critic system exhibits the matching behavior in the steady state of the learning even when the matching behavior is suboptimal. However, it is found that the matching behavior can earn approximately the same reward as the optimal one in many practical situations.     

Surface and deep learning processes in distance education: Synchronous versus asynchronous systems

Distance learning is different from regular learning in the classroom. One of the main factors which influence the effectiveness of the learning process is the interaction that exists between the teacher and the student. Our research indicates that different interactions have different effects.