How is Fuzzy Logic used in FLAME?
FLAME applies membership functions to various processing steps, such as the identification of elements, deconvolution, etc., rather than applying rigid constraints. The membership functions are constantly adapted as FLAME is being used for analysis. For example instead of saying an element is either present or not present, a membership probability is calculated, based on many parameters, with the result that the element may be classified as most probably present, or as unlikely. Deconvolution proceeds using whatever method is necessary, depending on the degree and complexity of overlap. Quantitation is dependent on analytical conditions. As FLAME is used, it learns from its input, and modifies its AI database as necessary, to compensate for the many inadequacies in the x-ray energy tables, absorption coefficients, and other physical parameters.
