Wavelet neural networks: A recent strategy for processing complex signals applications to chemistry

In the last three decades, Artificial Neural Networks (ANNs) have received increasing attention due to their wide and important applications in different areas of knowledge as adaptive tool for processing data. ANNs are, unlike traditional statistical techniques capable of identifying and simulating non-linear relationships in the information without any a priori assumptions about the data’s distribution properties. Furthermore, their abilities to learn, remember and compare, make them useful processing tools for many data interpretation tasks within analytical systems.
Nevertheless, the development of more complex analytical instruments and the need to cope with huge experimental data sets have demanded for new approaches in data analysis, leading to the development of advanced experimental designs and data processing methodologies based on novel computing paradigms, in order to tackle problems in areas such as calibration systems, pattern recognition, resolution and recovery of pure-components from overlapped spectra or mixtures.
This chapter describes the nature and function of Wavelet Neural Networks (WNNs), with clear advantages in topics such as feature selection, signal pre-processing, data meaning and optimization tasks in the treatment of chemical data. The chapter focuses on the last applications of WNNs in analytical chemistry as one of its most creative contributions from theoretical developments in mathematical science and artificial intelligence. Specifically, recent contributions from our laboratory showing their performance in electronic tongue applications will be outlined and commented.