Toward Self-Learning Model-Based EAS

Erik A. Meulman, Peter A.N. Bosman

Research output: Chapter in Book/Conference proceedings/Edited volumeConference contributionScientificpeer-review

Abstract

Model-based evolutionary algorithms (MBEAs) are praised for their broad applicability to black-box optimization problems. In practical applications however, they are mostly used to repeatedly optimize different instances of a single problem class, a setting in which specialized algorithms generally perform better. In this paper, we introduce the concept of a new type of MBEA that can automatically specialize its behavior to a given problem class using tabula rasa self-learning. For this, reinforcement learning is a naturally fitting paradigm. A proof-of-principle framework, called SL-ENDA, based on estimation of normal distribution algorithms in combination with reinforcement learning is defined. SL-ENDA uses an RL-agent to decide upon the next population mean while approaching the rest of the algorithm as the environment. A comparison of SL-ENDA to AMaLGaM and CMA-ES on unimodal noiseless functions shows mostly comparable performance and scalability to the broadly used and carefully manually crafted algorithms. This result, in combination with the inherent potential of self-learning model-based evolutionary algorithms with regard to specialization, opens the door to a new research direction with great potential impact on the field of model-based evolutionary algorithms.

Original languageEnglish
Title of host publicationGECCO'19
Subtitle of host publicationProceedings of the 2019 Genetic and Evolutionary Computation Conference Companion
Place of PublicationNew York
PublisherAssociation for Computing Machinery (ACM)
Pages1495-1503
Number of pages9
ISBN (Print)978-1-4503-6748-6
DOIs
Publication statusPublished - 2019
Event2019 Genetic and Evolutionary Computation Conference, GECCO 2019 - Prague, Czech Republic
Duration: 13 Jul 201917 Jul 2019

Conference

Conference2019 Genetic and Evolutionary Computation Conference, GECCO 2019
Country/TerritoryCzech Republic
CityPrague
Period13/07/1917/07/19

Keywords

  • Black-box optimization
  • Estimation of distribution algorithms
  • Machine learning
  • Reinforcement learning

Fingerprint

Dive into the research topics of 'Toward Self-Learning Model-Based EAS'. Together they form a unique fingerprint.

Cite this