2T-08
AN ENHANCED QUANTUM-INSPIRED GENETIC ALGORITHM WITH ARTIFICIAL ENTANGLEMENT: ANALYSIS OF ITS SEARCH BEHAVIOR AND POTENTIAL APPLICABILITY
○チーケン チョイ,ラック ターウォンマット(立命館大)
This paper describes a step-up from previously accepted paper at IEEE SSCI 2014 (Orlando, Florida) that further explains the algorithm's search behavior and a proposed improvement of artificial entanglement (AE) to solve an additional problem which is dubbed as the "Hamming Wall".

The "Hamming Wall" has been present since the use of bits to represent numbers as a weakness. This is a well-known problem where at certain point, in the case of a optimization phase such as mutation, the number sequence represented can no longer go further up or down due to the large hamming distance to jump to the next number. Hence, typical mutation strategies that attempts to jump to the next number sequence as part of the optimization often get stuck for a long time and unable to escape.

Therefore, this paper proposes a method that makes use of the artificial entanglement to resolve the issue as previously mentioned in prior paper, artificial entanglement has a wide degree of freedom where the initialization of the entangled Q-bits and the rotation gate can be freely defined.

Later section in this paper describes a brief experiment to demonstrate the applicability of AE and results are discussed.

footer 情報処理学会 セキュリティ プライバシーポリシー 倫理綱領 著作権について