Unbiased sensitivity analysis and pruning techniques in neural networks for surface ozone modelling

Abstract This paper presents the use of artificial neural networks (ANNs) for surface ozone modelling. Due to the usual non-linear nature of problems in ecology, the use of ANNs has proven to be a common practice in this field. Nevertheless, few efforts have been made to acquire knowledge about the problems by analysing the useful, but often complex, input–output mapping performed by these models. In fact, researchers are not only interested in accurate methods but also in understandable models. In the present paper, we propose a methodology to extract the governing rules of trained ANN which, in turn, yields simplified models by using unbiased sensitivity and pruning techniques. Our proposal has been evaluated in thousands of trained ANNs under different conditions to establish a relationship between present contaminants (or several atmospheric variables) and surface ozone concentrations. The technique presented has demonstrated to be unbiased and stable with regard to the interpretability of the models and the good results obtained.

[1]  Simon Haykin,et al.  Neural Networks: A Comprehensive Foundation , 1998 .

[2]  C L Mulchi,et al.  Wheat-kernel growth characteristics during exposure to chronic ozone pollution. , 1993, Environmental pollution.

[3]  S V Krupa,et al.  Use of tobacco cultivars as biological indicators of ambient ozone pollution: an analysis of exposure-response relationships. , 1993, Environmental pollution.

[4]  James E. Weber,et al.  ENVIRONMENTAL AUDITING: An Approach for Characterizing Tropospheric Ozone Risk to Forests , 1997, Environmental management.

[5]  Paul S. Bradley,et al.  Mathematical Programming for Data Mining: Formulations and Challenges , 1999, INFORMS J. Comput..

[6]  Fernando Gustavo Tomasel,et al.  Prediction of functional characteristics of ecosystems: a comparison of artificial neural networks and regression models , 1997 .

[7]  Jacek M. Zurada,et al.  Sensitivity analysis for minimization of input data dimension for feedforward neural network , 1994, Proceedings of IEEE International Symposium on Circuits and Systems - ISCAS '94.

[8]  Robert E. Uhrig,et al.  Using modular neural networks to monitor accident conditions in nuclear power plants , 1992, Defense, Security, and Sensing.

[9]  C. Willmott.  ON THE VALIDATION OF MODELS , 1981 .

[10]  Phil F. Culverhouse,et al.  Biological pattern recognition by neural networks , 1991 .

[11]  Patrick L. Kinney,et al.  The effects of ambient ozone on lung function in children: a reanalysis of six summer camp studies. , 1996 .

[12]  A. P. Altshuiler.  Association of oxidant episodes with warm stagnating anticyclones , 1978 .

[13]  Patrick Gallinari,et al.  FEATURE SELECTION WITH NEURAL NETWORKS , 1999 .

[14]  Sagar V. Krupa,et al.  Ambient ozone and adverse crop response : an evaluation of North American and european data at they relate to exposure indices and critical levels , 1995 .

[15]  Antonio J. Serrano,et al.  Dosage individualization of erythropoietin using a profile-dependent support vector regression , 2003, IEEE Transactions on Biomedical Engineering.

[16]  A. Comrie.  Comparing Neural Networks and Regression Models for Ozone Forecasting , 1997 .

[17]  Ian Colbeck,et al.  PHOTOCHEMICAL OZONE POLLUTION , 1985 .

[18]  I. Dimopoulos,et al.  Application of neural networks to modelling nonlinear relationships in ecology , 1996 .

[19]  K. Wilson,et al.  Linear stochastic models for forecasting daily maxima and hourly concentrations of air pollutants , 1975 .

[20]  Volker Tresp,et al.  Early Brain Damage , 1996, NIPS.

[21]  M. Kolehmainen,et al.  Neural networks and periodic components used in air quality forecasting , 2001 .

[22]  Giuseppe Nunnari,et al.  The application of neural techniques to the modelling of time-series of atmospheric pollution data , 1998 .

[23]  N. T. Kottegoda.  Linear stochastic models , 1980 .

[24]  M. Gevrey,et al.  Review and comparison of methods to study the contribution of variables in artificial neural network models , 2003 .

[25]  S V Krupa,et al.  Ambient ozone and crop loss: establishing a cause-effect relationship. , 1994, Environmental pollution.

[26]  William F. McDonnell,et al.  An air quality data analysis system for interrelating effects, standards, and needed source reductions: Part 11. A lognormal model relating human lung function decrease to O3 exposure. , 1991 .

[27]  Barry M. Wise,et al.  A comparison of neural networks, non-linear biased regression and a genetic algorithm for dynamic model identification , 1995 .

[28]  Babak Hassibi,et al.  Second Order Derivatives for Network Pruning: Optimal Brain Surgeon , 1992, NIPS.

[29]  M Lippmann,et al.  Effects of single- and multiday ozone exposures on respiratory function in active normal children. , 1991, Environmental research.

[30]  Nikhil R. Pal,et al.  Soft computing for feature analysis , 1999, Fuzzy Sets Syst..

[31]  Julian D. Olden,et al.  Illuminating the “black box”: a randomization approach for understanding variable contributions in artificial neural networks , 2002 .

[32]  Yann LeCun,et al.  Optimal Brain Damage , 1989, NIPS.

[33]  J. L. Carrasco-Rodriguez,et al.  Effective 1-day ahead prediction of hourly surface ozone concentrations in eastern Spain using linear models and neural networks , 2002 .

[34]  Paul Geladi,et al.  A comparison of modeling nonlinear systems with artificial neural networks and partial least squares , 1999 .

[35]  Greg Spellman,et al.  Analysing air pollution meteorology , 1998 .

[36]  Stephen Dorling,et al.  Statistical surface ozone models: an improved methodology to account for non-linear behaviour , 2000 .

[37]  Y. M. Tzeng,et al.  Distribution feeder loss analysis by using an artificial neural network , 1995 .

[38]  G. Soja,et al.  Ozone indices based on simple meteorological parameters: potentials and limitations of regression and neural network models , 1999 .

[39]  Wright-Patterson Afb,et al.  Feature Selection Using a Multilayer Perceptron , 1990 .

[40]  G. Spellman.  An application of artificial neural networks to the prediction of surface ozone concentrations in the United Kingdom , 1999 .