Spectral classification using pattern-recognition techniques. I. Feasibility with hydrogen as a model system

Pattern-recognition techniques are applied to the classification of atomic spectral transitions using hydrogen as a model system. Certain of the transitions have been randomly selected and treated as unclassified lines. Training of the known data at the 100% level has been achieved. Prediction of n₁ is 95.8%, while predictions of transitions of the type n₁, l₁→l₂ and n₁, l₁→l₂, J₁→J₂, are 83.4% and 66.67%, respectively. Using similar techniques, two unknown transitions in neutral atomic sodium have been classified.