Inductive functional programming using incremental program transformation
Artificial Intelligence
Fundamentals of algorithmics
Hi-index | 0.00 |
Program induction, where one or more parts of a potentially huge software system are automatically synthesized, is an emerging technology that will become more and more industrially useful as more computing power becomes available, for example in the form of computing grids. The scalability of program induction primarily depends on the combinatorial properties of the program transformations that are employed. In this paper, we improve the scalability of the ADATE automatic programming system by using a standard tool from combinatorial design theory, namely covering arrays, to more efficiently explore combinations of program transformations. The paper presents a series of 18 experiments which show that the covering array transformation algorithm is a highly useful supplement to the old ADATE transformations.