We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.NE

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Neural and Evolutionary Computing

Title: New Probabilistic-Dynamic Multi-Method Ensembles for Optimization based on the CRO-SL

Abstract: In this paper we propose new probabilistic and dynamic (adaptive) strategies to create multi-method ensembles based on the Coral Reefs Optimization with Substrate Layers (CRO-SL) algorithm. The CRO-SL is an evolutionary-based ensemble approach, able to combine different search procedures within a single population. In this work we discuss two different probabilistic strategies to improve the algorithm. First, we defined the Probabilistic CRO-SL (PCRO-SL), which substitutes the substrates in the CRO-SL population by {\em tags} associated with each individual. Each tag represents a different operator which will modify the individual in the reproduction phase. In each generation of the algorithm, the tags are randomly assigned to the individuals with a similar probability, obtaining this way an ensemble with a more intense change in the application of different operators to a given individual than the original CRO-SL. The second strategy discussed in this paper is the Dynamical Probabilistic CRO-SL (DPCRO-SL), in which the probability of tag assignment is modified during the evolution of the algorithm, depending on the quality of the solutions generated in each substrate. Thus, the best substrates in the search process will be assigned with a higher probability that those which showed a worse performance during the search. We test the performance of the proposed probabilistic and dynamic ensembles in different optimization problems, including benchmark functions and a real application of wind turbines layout optimization, comparing the results obtained with that of existing algorithms in the literature.
Comments: 18 pages, 6 figures, 5 tables
Subjects: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI); Optimization and Control (math.OC)
MSC classes: 68T01 and 68T20
Cite as: arXiv:2212.00742 [cs.NE]
  (or arXiv:2212.00742v2 [cs.NE] for this version)

Submission history

From: Jorge Pérez Aracil [view email]
[v1] Wed, 30 Nov 2022 15:05:19 GMT (1599kb,D)
[v2] Fri, 2 Dec 2022 09:25:24 GMT (1599kb,D)

Link back to: arXiv, form interface, contact.