On-line Learning and the Metrical Task System Problem

The problem of combining expert advice, studied extensively in the Computational Learning Theory literature, and the Metrical Task System (MTS) problem, studied extensively in the area of On-line Algorithms, contain a number of interesting similarities. In this paper we explore the relationship between these problems and show how algorithms designed for each can be used to achieve good bounds and new approaches for solving the other. Specific contributions of this paper include: An analysis of how two recent algorithms for the MTS problem can be applied to the problem of tracking the best expert in the decision-theoretic setting, providing good bounds and an approach of a much different flavor from the well-known multiplicative-update algorithms. An analysis showing how the standard randomized Weighted Majority (or Hedge) algorithm can be used for the problem of combining on-line algorithms on-line, giving much stronger guarantees than the results of Azar, Y., Broder, A., & Manasse, M. (1993). Proc ACM-SIAM Symposium on Discrete Algorithms (pp. 432–440) when the algorithms being combined occupy a state space of bounded diameter. A generalization of the above, showing how (a simplified version of) Herbster and Warmuth's weight-sharing algorithm can be applied to give a finely competitive bound for the uniform-space Metrical Task System problem. We also give a new, simpler algorithm for tracking experts, which unfortunately does not carry over to the MTS problem.Finally, we present an experimental comparison of how these algorithms perform on a process migration problem, a problem that combines aspects of both the experts-tracking and MTS formalisms.     

Modeling Robot “Psycho-Physical” State and Reactions – A New Option in Human–Robot Communication Part 1: Concept and Background

Man–machine communication had been recognized a long time ago as a significant issue in the implementation of automation. It influences the machine effectiveness through direct costs for operator training and through more or less comfortable working conditions. The solution for the increased effectiveness might be found in user-friendly human–machine interface. In robotics, the question of communication and its user-friendliness is becoming even more significant. It is no longer satisfactory that a communication can be called human–machine interface, since one must see robots as future collaborators, service workers, and probably personal helpers. So, a new way of communication is needed. This paper starts from the fact that a variety of unconventional ways of communication is being explored: speech communication, handwriting, and facial mimicry. Speech recognition and generation, apart from its potentially strong impact to human–robot interface, is out of the scope of this study, and the same holds for handwriting. The facial mimicry and the wider field of robot emotion lead to a new way of communication that can be called gestural language. We may say that our research matches such a wide notion. However, there is an important specificity of our approach. Previous works in this field looked at facial and gestural expressions of emotions and messages as actually voluntary activities. They were induced by some situations, but such situations had to be recognized after which the appropriate preprogrammed reaction followed. Thus, some expert system or other artificial intelligence system were applied. The current study explores the conditional reactions that are chemically based. Human psycho-physical states like fear, stress and especially fatigue follow from complex chemical processes. Such states cannot be voluntarily controlled. They are not expressed through some specific mimicry or body movement, but they influence all movements. Any motion intended to accomplish some task will be changed, thus sending a message to the neighborhood. This change can be observed, and so the message is received. For processes of this kind and their resembling, we prefer using the term modeling psycho-physical states. Among various states interesting for application in robots, the present study concentrates on fatigue, which is a good example for the initial research. The first part of the paper explains the concept and the background, while the second part provides numerical justification.     

Supporting dispatchability in schedules with consumable resources

Earlier work on scheduling by autonomous systems has demonstrated that schedules in the form of simple temporal networks, with intervals of values for possible event-times, can be made dispatchable, i.e. executable incrementally in real time with guarantees against failure due to unfortunate event-time selections. In this work we show how the property of dispatchability can be extended to networks that include constraints for consumable resources. We first determine conditions for insuring that resource use does not exceed capacity under dispatchable execution for a single sequence of activities, or bout, involving one resource. Then we show how to handle interactions between resource and temporal constraints to insure dispatchability, how to enhance flexibility of resource use under these conditions, and how to handle multiple bouts interspersed with instances of resource release. Finally, we consider methods for establishing the necessary dispatchability conditions during schedule creation (planning stage). The results demonstrate that flexible handling of resource use can be safely extended to the execution layer to provide more effective deployment of consumable resources.     

Virtual Connections and a Traffic Optimization Model for an ATM Network

The number of virtual connections in the nodal space of an ATM network of arbitrary structure and topology is computed by a method based on a new concept—a covering domain having a concrete physical meaning. The method is based on a network information sources—boundary switches model developed for an ATM transfer network by the entropy approach. Computations involve the solution of systems of linear equations. The optimization model used to compute the number of virtual connections in a many-category traffic in an ATM network component is useful in estimating the resource of nodal equipment and communication channels. The variable parameters of the model are the transmission bands for different traffic categories.     

InterViso: Dealing with the complexity of federated database access

Connectivity products are finally available to provide the highways between computers containing data. IBM has provided strong validation of the concept with their Information Warehouse. DBMS vendors are providing gateways into their products, and SQL is being retrofitted on many older DBMSs to make it easier to access data from standard 4GL products and application development systems. The next step needed for data integration is to provide (1) a common data dictionary with a conceptual schema across the data to mask the many differences that occur when databases are developed independently and (2) a server that can access and integrate the databases using information from the data dictionary. In this article, we discuss InterViso, one of the first commercial federated database products. InterViso is based on Mermaid, which was developed at SDC and Unisys (Templeton et al., 1987b). It provides a value added layer above connectivity products to handle views across databases, schema translation, and transaction management.     

Texture Analysis by Accurate Identification of Simple Markovian Models

A more accurate identification (estimation of parameters) of simple Markov-Gibbs random field models of images results in a better segmentation of specific multimodal images and realistic synthesis of some types of natural textures. Identification algorithms for segmentation are based in part on a novel modification of an unsupervised learning algorithm published first in Cybernetics and Systems Analysis (Kibernetika i Sistemnyi Analiz) almost four decades ago. A texture synthesis algorithm uses an identified model for selecting a characteristic geometric shape of and a placement grid for texture elements sampled from a training image utilized for the identification.__________Translated from Kibernetika i Sistemnyi Analiz, No. 1, pp. 37–49, January–February 2005.     

Accelerated simulation method for estimating the probability of functional failure in high-reliability systems

We consider a class of systems whose required performance is described by a stochastic process (t), t 0. An accelerated simulation method is proposed for estimating the probability that at some time instant t [0, T] the required performance exceeds the available performance. A numerical example is considered.Translated from Kibernetika i Sistemnyi Analiz, No. 4, pp. 30–41, July–August, 1991.     

Cardinality problems of compositions of morphisms and inverse morphisms

We show that we cannot effectively determine whether, for morphisms _i , _i ,card (u ₀ ^–1 ₁) =card(u ₀ ^–1 ₁) for all wordsu over the domain alphabets of the two given compositions. In contrast it is decidable for morphisms _i , _i and a regular setR whethercard(u _{0 1} ^–1 ) =card(u _{0 1} ^–1 ) for all wordsu inR. In order to prove the latter result we give a characterization of the multiplicity functions of simple finite automata by using cardinalities of compositions of the above form. Finally, we show that the above decidability result also holds when we consider rational functions rather than morphisms.     

A Review of:“ORTHOGONAL POLYNOMIALS FOR ENGINEERS AND PHYSICISTS”, by Petr Beckmann. The Golem Press,Boulder, Colorado, 1973. 280 pp.

This paper explores the problems that are associated with building a model (representation) of two systems which are central to the field of policy analysis. The systems are the Lockean and Dialectical Inquiring Systems (ISs). Given two or more belief systems, a Lockean IS will attempt to secure maximum agreement between them, whereas a Dialectical IS will attempt to secure maximum disagreement as the basis for forming a policy.

It is shown that a representation of these two systems necessitates a notion of probability assignment in which the law of conditional probabilities, p(A)p(B\A) = p(B)p(A\B), does not hold. Cross impact analysis provides such a notion of probability and hence is necessary to the representation of what happens as a decision-maker goes from a Lockean state to a Dialectical state. The movement from a Lockean to a Dialectical state is termed Conflict Production or Generation whereas the movement from a Dialectical to a Lockean state is termed Conflict Resolution or Reduction. It is argued that the representation of these two systems is basic to the foundation of a mathematical approach to policy analysis.     

Directional derivatives of the maximum function

Directional differentiability of the function (x) = sup{f(x, u), u U} is proved for a class of smooth functions f. The result is applied to study the directional differentiability of the function (x) = sup{f(x, y), y F(x)}, where F is a multivalued mapping.Translated from Kibernetika i Sistemnyi Analiz, No. 2, pp. 171–173, March–April, 1992.     

A General Framework for the Selection of World Coordinate Systems in Perspective and Catadioptric Imaging Applications

An imaging system with a single effective viewpoint is called a central projection system. The conventional perspective camera is an example of central projection system. A catadioptric realization of omnidirectional vision combines reflective surfaces with lenses. Catadioptric systems with an unique projection center are also examples of central projection systems. Whenever an image is acquired, points in 3D space are mapped into points in the 2D image plane. The image formation process represents a transformation from ³ to ², and mathematical models can be used to describe it. This paper discusses the definition of world coordinate systems that simplify the modeling of general central projection imaging. We show that an adequate choice of the world coordinate reference system can be highly advantageous. Such a choice does not imply that new information will be available in the images. Instead the geometric transformations will be represented in a common and more compact framework, while simultaneously enabling newer insights. The first part of the paper focuses on static imaging systems that include both perspective cameras and catadioptric systems. A systematic approach to select the world reference frame is presented. In particular we derive coordinate systems that satisfy two differential constraints (the compactness and the decoupling constraints). These coordinate systems have several advantages for the representation of the transformations between the 3D world and the image plane. The second part of the paper applies the derived mathematical framework to active tracking of moving targets. In applications of visual control of motion the relationship between motion in the scene and image motion must be established. In the case of active tracking of moving targets these relationships become more complex due to camera motion. Suitable world coordinate reference systems are defined for three distinct situations: perspective camera with planar translation motion, perspective camera with pan and tilt rotation motion, and catadioptric imaging system rotating around an axis going through the effective viewpoint and the camera center. Position and velocity equations relating image motion, camera motion and target 3D motion are derived and discussed. Control laws to perform active tracking of moving targets using visual information are established.     

The surface of language and humanities computing

Recent articles have noted that humanities computing techniques and methodologies remain marginal to mainstream literary scholarship. Mark Olsen's paper discusses this phenomenon and argues for large scale analyses of text databases that would incorporate a shift in theoretical orientation to include greater stress on intertextuality and sign theory. Part of Olsen's argument revolves on the need to move away from the syntactic and overt grammatical elements of textual language to more subtle semantics and meaning systems. While provocative and important, Olsen's stance remains rooted in literary theoretical constructs. Another level of language, the cognitive, offers equally interesting challenges for humanities computing, though the paradigms for this type of computer-based exploration are derived from disciplines traditionally removed from the humanities. The riddle, a nearly universal genre, offers a window onto some of the cognitive processes involved in deep level language function. By analyzing the riddling process, different methods of computational modelling can be inferred, suggesting new avenues for computing in the humanities.Charles Henry has a Ph.D. in Comparative Literature and is presently the Director of Libraries at Vassar College. His research includes non-literal aspects of language, and the cognitive processes involved in understanding symbolic language. Recent articles include The Image of Word,in Humanities and the Computer: New Directions, ed. D. Miall (Oxford, 1990), and Non-Literal Aspects of Language and Knowledge Structuring, inCybernetics and Systems Research '92, ed. R. Trappl (World Scientific Press, 1992).     

Stochastic Finite Learning of the Pattern Languages

The present paper proposes a new learning model—called stochastic finite learning—and shows the whole class of pattern languages to be learnable within this model.This main result is achieved by providing a new and improved average-case analysis of the Lange–Wiehagen (New Generation Computing, 8, 361–370) algorithm learning the class of all pattern languages in the limit from positive data. The complexity measure chosen is the total learning time, i.e., the overall time taken by the algorithm until convergence. The expectation of the total learning time is carefully analyzed and exponentially shrinking tail bounds for it are established for a large class of probability distributions. For every pattern containing k different variables it is shown that Lange and Wiehagen's algorithm possesses an expected total learning time of , where and are two easily computable parameters arising naturally from the underlying probability distributions, and E[] is the expected example string length.Finally, assuming a bit of domain knowledge concerning the underlying class of probability distributions, it is shown how to convert learning in the limit into stochastic finite learning.     

Toward Third Generation Threaded Discussions for Mobile Learning: Opportunities and Challenges for Ubiquitous Collaborative Environments

The mobile communication revolution has led to pervasive connectedness—as evidenced by the explosive growth of instant messaging in the home, and more recently, the enterprise–and, together with the convergence of mobile computing, provides a basis for extending collaborative environments toward truly ubiquitous immersion. Leveraging the true anytime/anywhere access afforded by mobile computing, it becomes possible to develop applications that not only are capable of responding to users whenever/wherever, on demand, but that also may actively seek out and engage users when the need arises. Thus, immersive environments need no longer be thought of strictly in terms of physical immersion with clearly discernable enter and exit events, but rather they may be extended, through mobile-enabled computing, toward ubiquity in terms of both time and space. Based on Media Synchronicity Theory, potential benefits are envisioned, particularly in the case of collaborative learning environments, from shortened response cycles and increased real time interaction opportunities. At the same time, a number of challenging issues must be addressed in designing such an environment to ensure user acceptance and to maximize realization of the potential. Third Generation (3G) Threaded Discussion has been conceptualized as an environment, well suited to mobile learning (m-learning) that could leverage mobile-enabled ubiquity to achieve a degree of extended immersion and thereby accrue the associated collaboration benefits. Exploring this conceptualization serves to help surface both the opportunities and the challenges associated with such environments and to identify promising design approaches, such as the use of intelligent agents.This revised version was published online in March 2005 with corrections to the cover date     

A database for linguists: Intelligent querying and increase of data

What do database (DB) querying and information retrieval imply for linguistics? What are data for the linguist? How can one envisage efficient access to data? I propose that DB querying in language sciences be designed linguistically and directly determined by linguistic data. Linguists from various backgrounds could then use a consensual query tool. An implemented data-directed DB querying system, which was developed for research on French interrogative structures, is presented in detail.Rachel Panckhurst teaches and researches in computational linguistics at the Université Paul Valéry-Montpellier III, France. Her research interests include natural language processing and information retrieval systems. She is currently devising a computer-assisted classification system for verbal subcategorization in French. She received her PhD (doctorat nouveau régime) in computational linguistics in December 1990 (Université Clermont II). Recent works include: Scatlex: une aide informatisée pour la construction d'entrées lexicales verbales, ICO Québec: Hiver 1993, pp. 61–67; Comment allier les besoins du linguiste et l'utilisation intelligente de bases de données? CIL '92 Proceedings, Québec, September 1993, pp 301–304.This article is a revised and extended English version of a paper in French, Comment allier les besoins du linguiste et l'utilisation intelligente de bases de données, Proceedings of the XVth International Congress of Linguists, Québec, August 1992, pp. 301–304, and was written during a post-doctoral fellowship (AUPELF-UREF) at the Centre d'ATO-CI, (Recherche et développement en linguistique computationnelle (RDLC)), Université du Québec à Montréal. The paper is adapted from part of Chapter 4 of a doctoral dissertation, Panckhurst, 1990. Many thanks are due to Jean-Marie Marandin who first encouraged me to write the paper and who read and gave constructive criticism on a previous version. I would also like to thank Sophie David and Claude Ricciardi Rigault (French version) and Fay Panckhurst (English version) for their helpful remarks. Finally, an anonymous reviewer provided constructive criticism, for which I am grateful. Any remaining errors are of course my own.     

Change Detection in Overhead Imagery Using Neural Networks

Identifying interesting changes from a sequence of overhead imagery—as opposed to clutter, lighting/seasonal changes, etc.—has been a problem for some time. Recent advances in data mining have greatly increased the size of datasets that can be attacked with pattern discovery methods. This paper presents a technique for using predictive modeling to identify unusual changes in images. Neural networks are trained to predict before and after pixel values for a sequence of images. These networks are then used to predict expected values for the same images used in training. Substantial differences between the expected and actual values represent an unusual change. Results are presented on both multispectral and panchromatic imagery.     

Managing the diffusion of organizational learning behavior

This paper addresses the need for organizations to manage the transformation from traditional hierarchical models to ‘learning organizations.’ We propose a five-stage methodology useful in the diffusion of behaviors associated with organizational learning (OL) theory. The stages of OL diffusion are (1) agenda-setting, (2) matching, (3) restructuring, (4) clarifying, and (5) routinizing. Each stage involves both managerial (structural) or member (cultural) influences on organizational memory (OM). Salient definitions are provided and the OM aspects and deliverables associated with each OL diffusion stage are discussed. This research provides a theoretically-driven approach to help change agents diffuse and realize the potential of OL behavior in the firm.

Enhancing the performance of an agent-based manufacturing system through learning and forecasting

Agent-based technology has been identified as an important approach for developing next generation manufacturing systems. One of the key techniques needed for implementing such advanced systems will be learning. This paper first discusses learning issues in agent-based manufacturing systems and reviews related approaches, then describes how to enhance the performance of an agent-based manufacturing system through learning from history (based on distributed case-based learning and reasoning) and learning from the future (through system forecasting simulation). Learning from history is used to enhance coordination capabilities by minimizing communication and processing overheads. Learning from the future is used to adjust promissory schedules through forecasting simulation, by taking into account the shop floor interactions, production and transportation time. Detailed learning and reasoning mechanisms are described and partial experimental results are presented.     

Rapid Concept Learning for Mobile Robots

Concept learning in robotics is an extremely challenging problem: sensory data is often high dimensional, and noisy due to specularities and other irregularities. In this paper, we investigate two general strategies to speed up learning, based on spatial decomposition of the sensory representation, and simultaneous learning of multiple classes using a shared structure. We study two concept learning scenarios: a hallway navigation problem, where the robot has to induce features such as opening or wall. The second task is recycling, where the robot has to learn to recognize objects, such as a trash can. We use a common underlying function approximator in both studies in the form of a feedforward neural network, with several hundred input units and multiple output units. Despite the high degree of freedom afforded by such an approximator, we show the two strategies provide sufficient bias to achieve rapid learning. We provide detailed experimental studies on an actual mobile robot called PAVLOV to illustrate the effectiveness of this approach.     

The blackboard model: a survey of its application

The need for co-operation and communication between Knowledge-Based Systems (KBSs) has prompted research into the field of Distributed Artificial Intelligence (DAI). A number of paradigms have been proposed — including the blackboard model.A de facto blackboard model is described which contains three components: the blackboard data structure, knowledge sources and a means for control. To enable comparison between existing applications, a set of attributes has been distilled from the model.Identification of three distinct groupings of current systems has led to the proposal of a taxonomy of blackboard systems. This consists of three generations of development: dedicated systems, generic shells and tool-based architectures.In light of this, an evaluation of the blackboard model is made, with respect to its significance to the field of DAI research.