An adaptive intelligent model for nucleotide sequence forecasting

The paper presents an adaptive retraining procedure that starts from a previously trained artificial neural network (ANN). The system is retrained to learn the evolution of a non-stationary sequence, without forgetting completely the previously learned data. The optimal ANN architecture is selected and the set of delayed input vectors is replaced with their principal components. The method is used for analyzing DNA genomic sequences.

[1]  I A Basheer,et al.  Artificial neural networks: fundamentals, computing, design, and application. , 2000, Journal of microbiological methods.

[2]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[3]  Dumitru-Iulian Nastac,et al.  Fast Retraining of Artificial Neural Networks , 2003, RSFDGrC.

[4]  M. Møller.  A Scaled Conjugate Gradient Algorithm for Fast Supervised Learning , 1990 .

[5]  Dumitru Iulian Nastac,et al.  Neuro-Adaptive Model for Financial Forecasting , 2007 .

[6]  Paul Dan Cristea,et al.  Large scale features in DNA genomic signals , 2003, Signal Process..

[7]  F. L. Lewis,et al.  Neural-network predictive control for nonlinear dynamic systems with time-delay , 2003, IEEE Trans. Neural Networks.

[8]  J. Edward Jackson,et al.  A User's Guide to Principal Components: Jackson/User's Guide to Principal Components , 2004 .

[9]  P D Cristea.  Conversion of nucleotides sequences into genomic signals , 2002, Journal of cellular and molecular medicine.

[10]  Martin Fodslette Møller,et al.  A scaled conjugate gradient algorithm for fast supervised learning , 1993, Neural Networks.

[11]  Marcin Wojnarski.  Prediction of product quality in glass manufacturing process using LTF-A neural network , 2003 .

[12]  I. Nastac,et al.  A retraining neural network technique for glass manufacturing data forecasting , 2004, 2004 IEEE International Joint Conference on Neural Networks (IEEE Cat. No.04CH37541).