Interactive Evolutionary Computation Using a Tabu Search Algorithm

• TAKENOUCHI Hiroshi

  Graduate School of Science and Engineering, Kansai University

• TOKUMARU Masataka

  Faculty of Engineering Science, Kansai University

• MURANAKA Noriaki

  Faculty of Engineering Science, Kansai University

Abstract

We present an Interactive Tabu Search (ITS) algorithm to reduce the evaluation load of Interactive Evolutionary Computation (IEC) users. Most previous IEC studies used an evaluation interface that required users to provide evaluation values for all candidate solutions. However, user's burden with such an evaluation interface is large. Therefore, we propose ITS where users choose the favorite candidate solution from the presented candidate solutions. Tabu Search (TS) is recognized as an optimization technique. ITS evaluation is simpler than Interactive Genetic Algorithm (IGA) evaluation, in which users provide evaluation values for all candidate solutions. Therefore, ITS is effective for reducing user evaluation load. We evaluated the performance of our proposed ITS and a Normal IGA (NIGA), which is a conventional 10-stage evaluation, using a numerical simulation with an evaluation agent that imitates human preferences (Kansei). In addition, we implemented an ITS evaluation for a running-shoes-design system and examined its effectiveness through an experiment with real users. The simulation results showed that the evolution performance of ITS is better than that of NIGA. In addition, we conducted an evaluation experiment with 21 subjects in their 20s to assess the effectiveness of these methods. The results showed that the satisfaction levels for the candidates generated by ITS and NIGA were approximately equal. Moreover, it was easier for test subjects to evaluate candidate solutions with ITS than with NIGA.

Journal

• IEICE Transactions on Information and Systems

  IEICE Transactions on Information and Systems E96.D (3), 673-680, 2013

  The Institute of Electronics, Information and Communication Engineers

Keywords

• interactive evolutionary computation
• user interface
• interactive tabu search

Details

• CRID: 1390001204379700224
• NII Article ID: 10031167456
• NII Book ID: AA10826272
• DOI: 10.1587/transinf.e96.d.673
• ISSN: 17451361
• ISSN: 09168532
• Web Site: https://www.jstage.jst.go.jp/article/transinf/E96.D/3/E96.D_673/_pdf
• Text Lang: en
• Data Source:
  • JaLC
  • Crossref
  • CiNii Articles
  • KAKEN
• Abstract License Flag: Disallowed