ALL libraries (COBIB.SI union bibliographic/catalogue database)
  • Neural-genetic system for modeling of antibiotic fermentation process
    Potočnik, Primož, 1969- ; Grabec, Igor
    Production of antibiotics is a complex secondary-metabolic process which is neither well understood nor analytically described. This paper presents empirical approach to modeling of antibiotic ... fermentation process. Several methods including neural networks, genetic algorithms and feature selection are combined with prior knowledge in the research methodology. The fermentation batch is regarded as one sample and the goal of modeling is to forecast the fermentation efficiency, which is defined by the highest achieved product concentration. The prior knowledge of experts from industry is utilized to extract the features for fermentation bacth characterization. The most representative features for our modeling purpose are determined in an optimization procedure with genetic algorithms and selected features are used to generate algorithms and selected features are used to generate the model. A linear model, a radial basis function neural network and hybrid linear-neural network model are applied for the model generation. Results show that the hybrid model is the most suitable for prediction of the fermentation efficiency. The presented modeling approach integrates the prior knowledge of experts with empirical information and represents a basis for the control of the fermentation process.
    Source: EIS 1998. Vol. 2, Neural networks (Str. 307-313)
    Type of material - conference contribution
    Publish date - 1998
    Language - english
    COBISS.SI-ID - 2453275

Shelf entry