Abstract
The problem of automation of neuro-fuzzy model synthesis on instance set is addressed. The method of instance selection for neuro-fuzzy model synthesis is proposed. It allows reducing the sample size, and decreasing the requirements to computer resources. The method also performs transformation of the original multi-dimensional coordinate set to the one-dimensional axis, which is also discretized to improve the data generalization properties. The software implementing proposed method is developed. The experiments were conducted to study the proposed method at the real problem solution. The results of experiments allow recommending proposed method for usage at practice.
Highlights
The neuro-fuzzy networks are a paradigm of computational intelligence widely used for building of diagnostic and recognizing models [1,2,3]
The characteristics of the original samples for the problem solution as well as the results of experiments on the proposed method investigation are given in the Table 1
The problem of autonomously partitioning the original sample into training and test samples that create instances of diagnostic and recognizing models have been addressed in the paper
Summary
The neuro-fuzzy networks are a paradigm of computational intelligence widely used for building of diagnostic and recognizing models [1,2,3]. The neuro-fuzzy network usually requires a training set of observations (instances) to build the model [3]. A number of problems faced with the need to process a large amount of available data which can not be loaded completely to the computer memory, as well as the fact that time of model building essentially depends on the training set volume. The actual problem is to reduce the volume of processed sample. It can be made through the allocation of training and test samples of smaller size from the available initial large sample. Known sampling methods are based on exhaustive search [4,5,6,7] and random search [5, 6, 8, 9]
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: 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.
