Behavioral code reuse

Abstract

To this point, our attempts to widen the evaluation bottleneck focused on defining alternative evaluation functions, which we conceptualize as search drivers in Chap. 9. However, an analysis of an execution record (Chap. 3), whether conducted with information-theoretic measures (Chap. 6) or machine learning algorithms (Chap. 7), also reveals information about the qualities of particular components of candidate solutions, i.e. subprograms. In this chapter, we elaborate on this observation and propose a means to harness its potential. We show how the detailed information available in an execution record together with behavioral evaluation enables (i) identification of useful components of programs, which can be then (ii) archived and (iii) reused by search operators. The following sections detail these stages as realized in [96].