Abstract
Several scholars have suggested using AI techniques to automatically develop algorithms, which is known as “hyper-heuristics”, to reduce the time and effort required in conventional methods. Although the Genetic Programming (GP) approach is the most popular hyper-heuristic approach used to generate dispatching rules to solve Job Shop Scheduling Problems (JSSPs), high computational requirements remain a major challenge for its wide applicability. Therefore, this paper proposes a mechanism to reduce the computational time needed to evaluate the solution quality of evolved rules. The proposed mechanism utilizes training data collected from the initial generation using a new representation to train a Support Vector Machine (SVM) classifier with a kernel of radial basis function. Then, in subsequent generations, the trained classifier is used to select the most promising (high-quality) rules for fitness assessment and discard low-performance ones. Consequently, only high-quality rules are evaluated, and the computational power that could have been used to evaluate poor rules is preserved. The performance of the proposed mechanism is analyzed using ten job shop instances from the literature, with respect to prediction accuracy and computational time. The results verify the effectiveness of the proposed approach in reducing the computational budget of the GP algorithm for JSSPs while achieving high training and testing accuracy.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.