Đang chuẩn bị liên kết để tải về tài liệu:
Vibrant search mechanism for numerical optimization functions

Không đóng trình duyệt đến khi xuất hiện nút TẢI XUỐNG Tải xuống

The results reveal that the proposed technique is able to orient the search mechanism toward desired segments of search space, which therefore significantly improves the overall search performance of HSA, especially for non-shifted optimization functions. | Journal of ICT, 17, No. 4 (October) 2018, pp: 679–702 How to cite this paper: Shambour, M., K. (2018).Vibrant search mechanism for numerical optimization functions. Journal of Information and Communication Technology. 17(4), 679-702. VIBRANT SEARCH MECHANISM FOR NUMERICAL OPTIMIZATION FUNCTIONS Moh’d Khaled Yousef Shambour Institute for Hajj and Umrah Research Umm Al-Qura University, Saudi Arabia myshambour@uqu.edu.sa ABSTRACT Recently, various variants of evolutionary algorithms have been offered to optimize the exploration and exploitation abilities of the search mechanism. Some of these variants still suffer from slow convergence rates around the optimal solution. In this paper, a novel heuristic technique is introduced to enhance the search capabilities of an algorithm, focusing on certain search spaces during evolution process. Then, employing a heuristic search mechanism that scans an entire space before determining the desired segment of that search space. The proposed method randomly updates the desired segment by monitoring the algorithm search performance levels on different search space divisions. The effectiveness of the proposed technique is assessed through harmony search algorithm (HSA). The performance of this mechanism is examined with several types of benchmark optimization functions, and the results are compared with those of the classic version and two variants of HSA. The experimental results demonstrate that the proposed technique achieves the lowest values (best results) in 80% of the non-shifted functions, whereas only 33.3% of total experimental cases are achieved within the shifted functions in a total of 30 problem dimensions. In 100 problem dimensions, 100% and 25% of the best results are reported for non-shifted and shifted functions, respectively. Received: 12 February 2018 Accepted: 18 September 2018 679 Published: 1 October 2018 Journal of ICT, 17, No. 4 (October) 2018, pp: 679–702 The results reveal that the proposed .