Scholarlink

<p>Meta-heuristics are utilized to handle challenging optimization problems, since conventional optimization techniques for such problems often fail or become stuck in the local optimum. Teaching-learning based optimization algorithm (TLBO) is a prominent meta-heuristic that mimics the teaching-learning process. It is initially employed to tackle unconstrained optimization problems. The main obstacle to meta-heuristics is premature convergence. The strategy of forming groups of students is introduced in the teaching-learning process for mutual discussions and joint projects. Along with some advantages, like increased creativity, diversity of ideas, access to new information, and more critical thinking, the group discussion strategy also has few disadvantages, such as disagreements over ideas, group size sensitivity, dependency on others, and lengthened the completion time. If properly implemented, the strategy can be a useful tool in the teaching-learning process. To increase an algorithm local search capabilities and to produce solutions of diverse nature, it is essential to have sufficient knowledge about the optimum solution with enough population diversity. Through group discussion sufficient information with diverse views about the solution of the problem is obtained. Therefore, in this work, the group discussion strategy is embedded to the learner phase of TLBO to enhance it. In the strategy, the group of randomly selected individuals correspond to each candidate solution is made to minimize the happening of premature convergence. This strategy depends on the group size parameter; for which the sensitivity analysis is also performed to obtain the right/optimal value. The suggested algorithm is known as GTLBO, and the group size parameter sensitivity analysis generates its four versions, referred to as GTLBO2–GTLBO5, where the digits 2–5 represent the group size value. The performance of the proposed algorithms is evaluated by the unconstrained benchmark problems of CEC 2017. The comparison of the obtained simulations’ results of the newly designed and some state-of-the-art algorithms on the tested problems exhibits that the suggested algorithms take the first, third, fourth, and fifth ranks in the top five ranks, which highlights the importance of the introduced strategy in the sense of enhancing TLBO.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing teaching learning based optimization algorithm through group discussion strategy for CEC 2017 benchmark problems

  • 期刊论文

  • 发表日期：

    2025年1月1日

查看全文
  • Muhammad Sagheer,
  • Muhammad Asif Jan,
  • Zahir Shah,
  • Wali Khan Mashwani,
  • Rashida Adeeb Khanum,
  • Meshal Shutaywi

摘要

Meta-heuristics are utilized to handle challenging optimization problems, since conventional optimization techniques for such problems often fail or become stuck in the local optimum. Teaching-learning based optimization algorithm (TLBO) is a prominent meta-heuristic that mimics the teaching-learning process. It is initially employed to tackle unconstrained optimization problems. The main obstacle to meta-heuristics is premature convergence. The strategy of forming groups of students is introduced in the teaching-learning process for mutual discussions and joint projects. Along with some advantages, like increased creativity, diversity of ideas, access to new information, and more critical thinking, the group discussion strategy also has few disadvantages, such as disagreements over ideas, group size sensitivity, dependency on others, and lengthened the completion time. If properly implemented, the strategy can be a useful tool in the teaching-learning process. To increase an algorithm local search capabilities and to produce solutions of diverse nature, it is essential to have sufficient knowledge about the optimum solution with enough population diversity. Through group discussion sufficient information with diverse views about the solution of the problem is obtained. Therefore, in this work, the group discussion strategy is embedded to the learner phase of TLBO to enhance it. In the strategy, the group of randomly selected individuals correspond to each candidate solution is made to minimize the happening of premature convergence. This strategy depends on the group size parameter; for which the sensitivity analysis is also performed to obtain the right/optimal value. The suggested algorithm is known as GTLBO, and the group size parameter sensitivity analysis generates its four versions, referred to as GTLBO2–GTLBO5, where the digits 2–5 represent the group size value. The performance of the proposed algorithms is evaluated by the unconstrained benchmark problems of CEC 2017. The comparison of the obtained simulations’ results of the newly designed and some state-of-the-art algorithms on the tested problems exhibits that the suggested algorithms take the first, third, fourth, and fifth ranks in the top five ranks, which highlights the importance of the introduced strategy in the sense of enhancing TLBO.