ICNN'97 TUTORIALS
SCHEDULE .......... ABSTRACTS ......... INSTRUCTORS .......... REGISTRATION
Attendees may attend a total of three tutorial workshops. Tutorials are on a first-come, first-served basis, and are subject to cancellation for inadequate pre-registration. ICNN 97 urges you to reserve your tutorial now to insure your place and to help avoid cancellation of this year's exciting tutorials.
Sunday, June 8
9:00 AM - 12:00 Noon
T1: Neural Networks for Consciousness.
J. G. Taylor. King's College, London
T2: Network ensembles and hybrid systems.
Joydeep Ghosh Univ. of Texas
1:30 - 4:30 PM
T3: Neuro-Fuzzy Recognition System: Concepts, Features and Feasibility.
Sankar K. Pal .Indian Statistical Institute
T4: Learning from Examples : from theory to practice.
Don R. Hush. University of New Mexico
6:00 - 9:00 PM
T5: Principles of Neurobiological Information Processing for Biology-Inspired Neural Computers.
Rolf Eckmiller. University of Bonn
T6: Hybrid Intelligent Information Systems - Models, Tools, Applications.
Nik Kasabov University of Otago Robert Kozma Tohoku University
TUTORIAL ABSTRACTS (Return to Top)
T1: Neural Networks for Consciousness
Presented by: Professor John G. Taylor, Institute for Medicine, Germany & King's College London, UK
Consciousness is now a topic of serious scientific analysis. New experimental results are pouring forth and neural network models are being developed to understand how the brain supports many information processing tasks. The same growth of experimental results and models is now occurring for consciousness, which is becoming recognised both as a complex of processes as well as being the crucial factor in the design of intelligent decision making systems. The tutorial will cover topics from: Difficulties; Nature; Past Approaches; New Windows; Global Principles; Local Emergence; Self; Emotions; Intelligence; Conclusions.
T2: Network Ensembles and Hybrid Systems
Presented by: Joydeep Ghosh, University of Texas at Austin, USA
Neural network ensembles and hybrid systems are increasingly being used for difficult multi-component problems for function approximation, pattern classification, and modeling/control of dynamic systems. This tutorial will first summarize recent results on how to properly combine multiple networks for regression or classification. Practical aspects of using input space partitioning methods such as mixture of experts will then be discussed. Finally, we will consider some hybrid systems where the complimentary nature of symbolic, connectionist and fuzzy techniques are exploited for better system performance and interpretation of results. Several case studies will be presented. For more details see www.lans.ece.utexas.edu.
T3: Neuro-Fuzzy Recognition System: Concepts, Features and Feasibility
Presented by: Sankar K. Pal, Indian Statistical Institute, India
Basic characteristics and features of fuzzy set theory and artificial neural networks, and their relevance to pattern recognition and image processing problems are, first of all, described. Then it categorizes the various ways of integrations of these two technologies, so far made, under the heading neuro-fuzzy computing along with generic and application specific advantages of the fusion, and provides some examples on such integration including methods of linguistic rule generation. Finally, the utility of genetic algorithms in neuro-fuzzy framework, their collective role as soft computing paradigm and the relation to future generation computing systems are explained.
T4: Learning from Examples: From Theory to Practice
Presented by: Dr. Don R. Hush, University of New Mexico, USA
This tutorial provides an overview of the problem of learning from examples. Emphasis is placed on fundamental limitations in three areas: approximation, estimation and computation. Each of these is compared and contrasted in situations where the problem is one of regression versus pattern classification, parametric versus nonparametric, and linear versus nonlinear.
General methods for improving generalization and computation speed are discussed, and practical examples are used to illustrate these methods. All attendees should have either a graduate degree, or be working on one, in neural networks or a closely related scientific discipline.
T5: Principles of Neurobiological Information Processing for Biology-Inspired Neural Computers
Presented by: Rolf Eckmiller, University of Bonn, Germany
The similarity between brain functions and most available neural computers (e.g. MLPs or RBF nets) is still minimal.
Specifically, brain functions are not synchronous or MLP-like and do not function similar to McCulloch-Pitts neurons.
Accordingly, the development of Biology-Inspired Neural Computers requires the understanding and incorporation of the essential structural and functional principles of biological neural systems as they emerge from neuroscience and computational neuroscience.
T6: Hybrid Intelligent Information Systems - Models, Tools, Applications
Presented by: Prof. N.K. Kasabov and Dr. Robert Kozma, University of Otago, New Zealand & Prof. Ron Sun, University of Alabama, USA
The Tutorial compares various artificial intelligence paradigms in combination with fuzzy systems, neural networks, evolutionary and chaotic models. Different ways of integrating them into intelligent hybrid informationsystems are presented.
Knowledge plays a key role through the Tutorial. Dynamic systems models and chaotic processes are utilized for the analysis of spatio-temporal evolution of hybrid systems. Chaotic self-organization generates hierarchical structures in a natural way that facilitates dynamic adaptation under changing external conditions. Software and hardware means to achieve these goals are demonstrated, including speech recognition, chaotic time series prediction, and adaptive control. A hybrid software environment FuzzyCope/2 is also demonstrated.
Title : Neuro-Fuzzy Recognition System: Concepts, Features and Feasibility.
Instructor : Sankar K. Pal.
Head, Machine Intelligence Unit,
Indian Statistical Institute
203, B. T. Road,
Calcutta 700 035. India.
Tel(O) : (91)-(33)-556-8085. Tel(R) : (91)-(33)-557-8030. Fax : (91)-(33)-556-6680/6925.
E-mail : sankar@isical.ernet.in (Return to Top)
Title : Network ensembles and hybrid systems.
Instructor : Joydeep Ghosh
Department of Electrical & Computer Engr.
Engineering Sciences Bldg. (ENS) 516
Univ. of Texas, Austin, TX-78712-1084. US
Tel : (512)-471-8980 ; Fax : (512)-471-5907.
Email : ghosh@ece.utexas.edu (Return to Top)
Title : Searching for Consciousness.
Instructor : J. G. Taylor.
(King's College, London)
Institut fur Medizin
Forshungszentrum Juelich
D-52428, Juelich GERMANY
fax: 0049-2461-61 28 20
e-mail: taylor@medicom03.ime.kfa-juelich.de
(udah057@bay.cc.kcl.ac.uk) (Return to Top)
Title : Principles of Neurobiological Information Processing for Biology-Inspired Neural Computers.
Instructor : Rolf Eckmiller.
Department of Computer Science,
University of Bonn,
Roemerstr. 164
D-53117 Bonn, F.R. Germany
Tel : 49-228-734-422 ; Fax : 49-228-734-425.
Email : eckmiller@nero.uni-bonn.de (Return to Top)
Title : Learning from Examples : from theory to practice.
Instructor : Don R. Hush.
EECE Department
University of New Mexico
Albuquerque, NM 87131.
Tel : 505-277-1611. Fax : 505-277-1439.
Email : hush@eece.unm.edu (Return to Top)
Title : Hybrid Intelligent Information Systems - Models, Tools, Applications.
Instructors : Nik Kasabov
Department of Information Sciences
University of Otago,
Dunedin, New Zealand.
Tel : +64 3 479 8319 ; Fax : +64 3 479 8311.
Email : nkasabov@otago.ac.nz (Return to Top)
Robert Kozma
Department of Quantum Science & Engineering
Laboratory Machine Intelligence/Measurement & Instrum.
Tohoku University, Sendai 980-77, Japan.
Tel : +81-22-217-7906 ; Fax : +81-22-217-7900.
E-mail : kozma@mine1.nucle.tohoku.ac.jp (Return to Top)
TUTORIAL CHAIR (Return to Top)
Prof. John Yen