Tabu Thresholding: Improved Search by Nonmonotonic Trajectories
Tabu Thresholding: Improved Search by Nonmonotonic Trajectories
复制标题
DOI：
10.1287/ijoc.7.4.426
复制
发表时间：
1995-11
期刊：
影响因子：
--
通讯作者：
F. Glover
中科院分区：
文献类型：
--
作者：
F. Glover
There is an appeal to methods like simulated annealing and threshold acceptance that operate by imposing a monotonically declining ceiling on objective function levels or degrees of disimprovement (treated probabilistically or deterministically). An alternative framework, embodied in tabu search, instead advocates a nonmonotonic form of control, keyed not only to the objective function but to other elements such as values of variables, direction of search, and levels of feasibility and infeasibility. This creates a more flexible search behavior and joins naturally with the use of memory-based strategies that are the hallmark of tabu search approaches. Embodied particularly in the strategic oscillation component of tabu search, this nonmonotonic control has been shown in a variety of studies to yield outcomes superior to those of simulated annealing and threshold acceptance. The question arises whether such an approach offers a sufficiently rich source of search trajectories to be relied upon as a primary ...
