Applications of Pattern Recognition to Chemistry

Abstract

“Pattern recognition” is difficult to define because various terms have come to have different meanings to various groups. References to “pattern recognition,” “learning machines,” “machine decisions,” “artificial intelligence,” “empirical decision processes,” and other names are scattered throughout the literature. Different groups of users, such as electrical engineers, physical scientists, neurobiologists, linguists, and psychologists, have adopted different phrases and have attached their own intrepretations to them. Even in chemistry conflicts in usage occur, as evidenced by the early use of “pattern recognition” in mass spectrometry to mean the direct comparison of one mass spectrum to a library of standards as an attempt to identify compounds. Indeed, one might say that all attempts to correlate information in any form amount to the recognition of patterns that represent that information. The problem becomes one of restricting the definition so that it can have some useful connotation.