An experimental analysis of a GP hyperheuristic approach for evolving low-cost heuristics for profile reductions

Abstract

Researchers used graph-theory approaches to design the state-of-theart low-cost heuristics for profile reduction. This paper evolves and selects four low-cost heuristics for profile reduction using a genetic programming hyperheuristic approach. This paper evaluates the resulting heuristics for profile reduction from the genetic programming hyperheuristic approach in two application areas against the low-cost heuristics for solving the problem. The results obtained on a set of standard benchmark matrices taken from the SuiteSparse sparse matrix collection indicate that the resulting heuristics from the genetic programming hyperheuristic approach does not compare favorably with a high-quality heuristics for profile reduction.

Low-cost heuristics for matrix bandwidth reduction combined with a Hill-Climbing strategy

“…Recently, Zhu et al [32] proposed a genetic programming hyper-heuristic approach for the multi-skill resource-constrained project scheduling problem. Along the same lines, the authors of the present study proposed a genetic programming hyper-heuristic strategy for evolving low-cost heuristics for profile reductions [27]. Tian et al [28] proposed an ACO-based hyper-heuristic with dynamic decision blocks for intercell scheduling.…”