### Wednesday 24th April 1996

- 8H30
*Registration* - 9H00
*Opening session*#### Session 1: Learning and generalization 1

Chairman: H. Paugam-Moisy (E.N.S. Lyon, France) - 9H10 "Synaptic efficiency modulations for context integration:
The meta ODWE architecture"

N. Pican

INRIA Lorraine (France) - 9H30 "Using a Meta Neural Network for RPROP parameter
adaptation"

C. McCormack

University College Cork (Ireland) - 9H50 "On unlearnable problems -or- A model for premature
saturation in backpropagation learning"

C. Goerick, W. von Seelen

Ruhr-Universität Bochum (Germany) - 10H10 "Regulated Activation Weights Neural Network (RAWN)"

H.A.B. te Braake*, H.J.L. van Can*, G. van Straten**, H.B. Verbruggen*

*Delft Univ. of Technology (The Netherlands)

** Wageningen Agricultural Univ. (The Netherlands) - 10H30 "Praticing Q-learning"

J. Bruske, I. Ahrns, G. Sommer

Christian Albrechts Univ. of Kiel (Germany) - 10H50
*Coffee break*#### Session 2: Recurrent models

Chairman: W. Duch (Nicholas Copernicus Univ., Poland) - 11H30 "Adaptative time constants improve the dynamic
features of recurrent neural networks"

J.-P. Draye*, D. Pavisic*, G. Cheron**, G. Libert*

* Fac. Polyt. Mons (Belgium), **Univ. of Brussels (Belgium) - 11H50 "An adaptive technique for pattern recognition
by the random neural network"

M. Mokhtari*,**, H. Akdag**,***

*IUT de Reims, **Univ. Paris V, ***Univ. Paris VI (France) - 12H10 "Negative initial weights improve learning in
recurrent neural networks"

D. Pavisic, J.-P. Draye, R. Teran, G. Calderon, G. Cheron, G. Libert

Fac. Polyt. Mons (Belgium) - 12H30
*Lunch*#### Session 3: Fuzzy neural networks

Chairman: M. Verleysen (Univ. Cath. de Louvain, Belgium) - 14H00 "The extraction of Sugeno fuzzy rules from neural
networks"

A.L. Cechin, U. Epperlein, W. Rosenstiel, B. Koppenhoefer

Univ. Tübingen (Germany) - 14H20 "Neural versus neurofuzzy systems for credit approval"

S. Piramuthu

Univ. of Florida (USA) - 14H40
*Invited paper:*

B. Fritzke (Ruhr-Universität Bochum, Germany)

"Growing self-organizing networks - Why ?" - 15H25
*coffee break*#### Session 4: Self-organizing maps

Chairman: B. Fritzke (Ruhr Universität Bochum, Germany) - 16H00 "Identification of gait patterns with self-organizing
maps based on ground reaction force"

M. Köhle, D. Merkl

Vienna Univ. of Technology (Austria) - 16H20 "A self-organizing map for analysis of high-dimensional
feature spaces with clusters of highly differing feature density"

S. Schünemann, B. Michaelis

Otto-von-Guericke Univ. Magdeburg (Germany) - 16H40 "Quantization vs Organization in the Kohonen S.O.M."

J.-C. Fort*, G. Pagès**

*Univ. Nancy I & Univ. Paris I (France)

**Univ. Pierre et Marie Curie & Univ. Paris XII (France) - 17H00 "On global self-organizing maps"

W. Duch, A. Naud

Nicholas Copernicus Univ. (Poland) - 17H20 "On the critical points of the 1-dimensional competitive
learning vector quantization algorithm"

D. Lamberton*, G. Pagès**

*Univ. Marne-la-Vallée, **Univ. Pierre et Marie Curie & Univ. Paris XII (France) - 17H40 "A Kohonen map representation to avoid misleading
interpretations"

M. Cottrell*, E. de Bodt**

*Univ. Paris I (France), Univ. Cat. Louvain (Belgium)

### Thursday 25th April 1996

- 9H00 "FlexNet - A flexible neural network construction
algorithm"

K. Mohraz, P. Protzel

FORWISS, Germany - 9H20 "Incremental category learning in a real world
artifact using growing dynamic cell structures"

C. Scheier

Univ. of Zurich (Switzerland) - 9H40 "Towards constructive and destructive dynamic network
configuration"

S. Wermter, M. Meurer

Univ. of Hamburg (Germany) - 10H00 "Combining sigmoids and radial basis functions
in evolutive neural architectures"

R. Chentouf, C. Jutten

I.N.P. Grenoble (France) - 10H20
*Coffee break* - 11H00
*Invited paper:*

N. Franceschini (CNRS Marseilles, France)

"Biologically inspired eye movements for visually guided navigation of mobile robots"

F. Mura, N. Martin, N. Franceschini#### Session 6: Classification

Chairman: J. Vandewalle (Kat. Univ. Leuven, Belgium) - 11H45 "A novel two-layer neural network classifier"

G. Qiu

Univ. of Derby (United Kingdom) - 12H05 "Representation of obstacles in a neural network
based classifier system"

N.R. Ball

Univ. of Cambridge (United Kingdom) - 12H25 "Investigating lexical access using neural nets"

M. Patel

Macquarie Univ. (Australia) - 12H45
*Lunch*#### Session 7: Mathematical aspects of neural networks

Chairman: J.C. Fort

(Univ. Nancy I & Univ. Paris I, France) - 14H00 "Rates of approximation of real-valued boolean
functions by neural networks"

K. Hlavackova, V. Kurkova

Acad. of Sciences (Czech Republic) - 14H20 "Bounds on the degree of high order binary perceptrons"

E. Mayoraz

IDIAP (Switzerland) - 14H40 "A fast Bayesian algorithm for Boolean functions
synthesis by means of perceptron networks"

A. Catala Mallofre*, J. Aguilar-Martin*,**, B. Morcego Seix*, N. Piera i Carrete*

*Univ. Polit. Catalunya (Spain), **LAAS-CNRS (France) - 15H00 "Accomodating relevance in neural networks"

H. Wang, D. Bell

Univ. of Ulster at Jordanstown (United Kingdom) - 15H20
*Coffee break*#### Session 8: Natural and artificial vision

Chairman: N. Franceschini (CNRS Marseilles, France) - 16H00 "Neural network for visual contrast detection"

E. Littmann*, H. Neumann*, L. Pessoa**

*Univ. of Ulm (Germany), **Univ. Federale Rio (Brazil) - 16H20 "Application of high-order Boltzmann machines
in OCR"

A. de la Hera, M. Grana, A. D'Anjou, F.X. Albizuri

UPV San Sebastian (Spain) - 16H40 "Simulation of an inner plexiform layer neural
circuit in vertebrate retina leads to sustained and transient excitation"

G. Maguire, X. Yang

Univ. of Texas (USA) - 17H00 "Analysis of visual information by receptivc field
dynamics"

C. Born

Ruhr-Univ. Bochum (Germany) - 17H20 "Neurotransmitterdynamics in a model of a movement
detecting visual system"

H.A.K. Mastebroek

Groningen Univ. (The Netherlands) - 20H00
*Conference dinner*

#### Session 5: Incremental learning

Chairman: T.W.S. Chow (City University of Hong Kong)

### Friday 26th April 1996

- 9H00 "Recurrent least square learning for quasi-parallel
principal component analysis"

W. Kasprzak, A. Cichocki

RIKEN Frontier Research (Japan) - 9H20 "Understanding data by neural and statistical tools"

A. Guérin-Dugué, C. Avilez-Cruz, P.M. Palagi

I.N.P. Grenoble (France) - 9H40 "Error rate estimation via cross-validation and
learning curve theory"

A. Varfis, L. Corleto

CEC Joint Research Center Ispra (Italy) - 10H00 "Maximum covariance method for weight initialization
of multilayer perceptron network"

M. Lehtokangas*, P. Korpisaari**, K. Kaski**

*Nokia Research Center, **Tampere Univ. of Tech. (Finland) - 10H20
*Coffee break* - 11H00
*Invited paper*

J. Karhunen (Helsinki Univ. of Technology, Finland)

"Neural approaches to independent component analysis and source separation"#### Session 10: Learning and generalization 2

Chairman: E. Mayoraz (IDIAP Martigny, Switzerland) - 11H40 "Constraining of weights using regularities"

J.N. Kok*, E. Marchiori*,**, M. Marchiori***, C. Rossi****

*Univ. of Leiden (The Netherlands), **CWI (The Netherlands), ***Univ. of Padova (Italy), ****Univ. of Venezia (Italy) - 12H00 "Regularization and neural computation: application
to aerial images analysis"

E. Schaeffer, P. Bourret, S. Montrozier

ONERA-CERT (France) - 12H20
*Lunch*#### Session 10: Learning and generalization 2 (continued)

Chairman: E. Mayoraz (IDIAP Martigny, Switzerland) - 14H00 "An algorithm for training multilayer networks
on non-numerical data"

W. Kowalczyk

Vrije Univ. Amsterdam (The Netherlands) - 14H20 "A correlation-based network for real-time processing"

J. Ngole

Uppsala Univ. (Sweden) - 14H40 "Evolving neural network learning behaviours with
set-based chromosomes"

S.M. Lucas

Univ. of Essex (United Kingdom) - 15H00
*coffee break*#### Session 11: Prediction

Chairman: M. Cottrell (Univ. Paris I, France) - 15H30 "Neural network application: rainfall forecasting
system in Hong Kong"

T.W.S. Chow, S. Cho

City Univ. of Hong-Kong - 15H50 "Prediction of dynamical systems with composition
networks"

Y. Moreau, J. Vandewalle

Kat. Univ. Leuven (Belgium) - 16H10 "Fast signal recognition and detection using ART1
neural networks and nonlinear preprocessing units based on time delay embeddings"

R. Dogaru, A.T. Murgan, C. Comaniciu

Polyt. Univ. of Bucharest (Romania) - 16H30 "An analysis of the metric structure of the weight
space of feedforward networks and its application to time series modeling
and prediction"

A. Ossen, S.M. Rüger

Tech. Univ. Berlin (Germany) - 16H50 "Time series prediction using neural networks
and its application to artificial human walking"

R.S. Venema, A. Ypma, J.A.G. Nijhuis, L. Spaanenburg

Rijksuniv. Groningen (The Netherlands)