ALL libraries (COBIB.SI union bibliographic/catalogue database)
  • Application of genetic and neural network algorithms in vacuum science
    Belič, Igor, 1960- ; Irmančnik-Belič, Lidija
    Vacuum science evolves very rapidly, but vacuum science development is highly dependant on successful application of new mathematical methods for analysis of measurements. The mathematical methods ... together with computer science can provide further boost to the cacuum science. The paper describes new methods studied through several years for the mass spectra analysis although it attempts to be general and usable in any branch of measurements analysis (spectral and others). The ability of neural networks to develop the knowledge with learning is very suitable for all kind of pattern recognition problems, where exact methematical models are not known, or where the models are quite well known but several parameters are omitted due to complexity of the model. Practical use of the classical methematical (statistical) tools isthus very much limited to the problems where the exact models are known. The neural networks applications in mass spectrometry were thoroughly studied and the findings were already published. This paper focuses on new aspects - with introduction of genetic algorithms to enhance the already well known neural network algorithms. Genetic algorithms provide a very powerful improvement to the neural networks. Due to the importance and impact of the new mathematical and computer tools on the vacuum science, authors propose the formation of a new group within the vacuum science framework taht will concentrate the efforts on applications of these methods in the vacuum science.
    Source: Abstract book (Str. 407)
    Type of material - conference contribution
    Publish date - 1998
    Language - english
    COBISS.SI-ID - 19930629