Artificial intelligence-based pattern recognition

One of the earliest applications of techniques of artificial intelligence (a term used here in its broadest sense) in decisionmaking was the reduction of spectral data to their structural implications, information that is usually expressed in terms of substructures predicted to be present or absent in the unknown. The actual techniques used generally fall into one of three classes library search, pattern recognition, and knowledge-based systems. However, the boundaries between these classes are not sharp. Regardless of the technique selected, it is important that the spectral data should be diagnostic for the type of structural features the user needs to identify. 

---