The linear learning machine method was applied to the determination of cadmium, lead and thallium down to 10-8 M by anodic stripping voltammetry at a hanging mercury drop electrode. With a total of three trained multicategory classifiers, concentrations of Cd, Pb and Tl could be predicted with an accuracy of ±10%. The classifiers were trained with the use of least-squares minimization. Numerical problems in the data matrix inversion were overcome by using singular value decomposition.
Bos, M., & Jasink, G. (1978). The learning machine in quantitative chemical analysis : Part I. Anodic Stripping Voltammetry of Cadmium, Lead and Thallium. Analytica chimica acta, 103(2), 151-165. https://doi.org/10.1016/S0003-2670(01)84035-6