Title of article :
The NeuroDeveloper®: a tool for modular neural classification of spectroscopic data
Author/Authors :
Udelhoven، نويسنده , , Thomas and Novozhilov، نويسنده , , Mark and Schmitt، نويسنده , , Jürgen، نويسنده ,
Issue Information :
دوفصلنامه با شماره پیاپی سال 2003
Pages :
8
From page :
219
To page :
226
Abstract :
The NeuroDeveloper® is a high-performance software tool for the classification of spectroscopic data using artificial neural networks (ANNs). The software consists of four specialized modules, each constructed for one step within the classification procedure. The FeatureDeveloper offers a user-friendly visualization of the raw-spectra, uni- or multivariate methods for wavelength selection and compression, as well as standard tools for spectra preprocessing. The NeuroSimulator performs the neural network training. The identification of a suited network architecture is supported by an automation tool. The development of modular and hierarchically organized neural networks of any degree of complexity is done in the ModuleDeveloper. Finally, the evaluation of new data by trained ANNs is achieved by the Classification module with a complete documentation via printout and log-files.
Keywords :
NEURAL NETWORKS , Modular ANNs , Classification , feature selection , Spectroscopic data
Journal title :
Chemometrics and Intelligent Laboratory Systems
Serial Year :
2003
Journal title :
Chemometrics and Intelligent Laboratory Systems
Record number :
1460753
Link To Document :
بازگشت