An attribute difference revision method in case-based reasoning and its application. (October 2017)

• Record Type: Journal Article
• Authors: Yan, Aijun; Zhang, Kuanhong; Yu, Yuanhang; Wang, Pu
• Is Part Of: Engineering applications of artificial intelligence. Volume 65(2017:May)
• Page Start: 212
• Page End: 219
• Publication Date: 2017-10
• Subjects: Case-based reasoning -- Attribute difference -- Case revision -- Support vector regression; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2017.07.015
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: To improve the learning performance of the revision stage in case-based reasoning (CBR), an attribute difference revision method (ADR) is proposed in this paper. First, the suggested solution of the target case is obtained through the case retrieval and case reuse; then, the revision value of the suggested solution and output results of the CBR model are obtained by using the support vector regression (SVR) model, which is based on the difference between the target case and similar cases; finally, the target case and its correct solutions are stored. Experiments and applications shows that the ADR method is effective and the fitting error of the ADR-based CBR (ADRCBR) model is significantly lower than other typical regression methods, indicating that ADR can improve the learning performance of the CBR model and has the advantage of application. Highlights: An attribute difference revision (ADR) based CBR model (ADRCBR) is proposed. The revision value is obtained by the SVR in the case revision phase. The ADR method can improve the learning performance of the CBR model. The fitting error of the ADRCBR model is significantly lower than other typical regression methods.