Visualization of Neural Net Evolution

Genetic identification of models of dynamical systems is becoming a well stablished research field. Nowadays it is hard to obtain more precise numerical results than state of the art methods, but, in our oppinion, there is still room to improve the understandability of genetically induced models. In this paper it is proposed a method that focuses in the comprehensibility of the final model, while keeping most of the numerical precision of former studies. The main innovation in this work is centered in the concept of \understandable" system. We do not use state space designed, rule based models, but z-transform based models, comprising linear, discrete dynamical models of first or second order and memoriless nonlinear elements (saturation, dead zone or other nonlinear gains.) This way, we provide control engineers with their prefered representation in moderate to complex models, and facilitate the task of designing control systems for these processes.

[1]  Brian D. Ripley,et al.  Pattern Recognition and Neural Networks , 1996 .

[2]  James P. Rice,et al.  Finding an Impulse Response Function Using Genetic Programming , 1993, 1993 American Control Conference.

[3]  Kurt Dirk Bettenhausen,et al.  Signal path oriented approach for generation of dynamic process models , 1996 .

[4]  Les M. Howard,et al.  The GA-P: A Genetic Algorithm and Genetic Programming Hybrid , 1995, IEEE Expert.

[5]  David J. Murray-Smith,et al.  Nonlinear model structure identification using genetic programming , 1998 .

[6]  Peter Marenbach,et al.  Using prior knowledge and obtaining process insight in data based modelling of bioprocesses , 1998 .

[7]  Paul Horton,et al.  Better Prediction of Protein Cellular Localization Sites with the it k Nearest Neighbors Classifier , 1997, ISMB.

[8]  Francisco Herrera,et al.  A proposal for improving the accuracy of linguistic modeling , 2000, IEEE Trans. Fuzzy Syst..

[9]  Hitoshi Iba,et al.  System Identification using Structured Genetic Algorithms , 1993, ICGA.

[10]  Christopher M. Bishop,et al.  Neural networks for pattern recognition , 1995 .

[11]  Jeanny Hérault,et al.  Curvilinear component analysis: a self-organizing neural network for nonlinear mapping of data sets , 1997, IEEE Trans. Neural Networks.

[12]  Arthur Flexer,et al.  On the use of self-organizing maps for clustering and visualization , 1999, Intell. Data Anal..

[13]  Luciano Sánchez Ramos,et al.  Niching scheme for steady state GA-P and its application to fuzzy rule based classifiers induction , 2000 .

[14]  Michel Verleysen,et al.  Curvilinear Distance Analysis versus Isomap , 2002, ESANN.

[15]  John W. Sammon,et al.  A Nonlinear Mapping for Data Structure Analysis , 1969, IEEE Transactions on Computers.

[16]  L Tsogo,et al.  Multidimensional Scaling Methods for Many-Object Sets: A Review , 2000, Multivariate behavioral research.

[17]  S T Roweis,et al.  Nonlinear dimensionality reduction by locally linear embedding. , 2000, Science.