Concept

In machine learning theory and artificial intelligence, a concept  over a domain  is a Boolean function . A collection of concepts is called a concept class.

In context-specific applications, concepts are usually thought to assign either a "positive" or "negative" outcome (corresponding to range values of 1 or 0, respectively) to each element  of the domain . In that way, concepts are the fundamental component of learning theory.

REFERENCES:

Shashua, A. "Lecture 11: PAC II." 2009. http://www.cs.huji.ac.il/~shashua/papers/class11-PAC2.pdf.