Design of Soft-max Radial Basis Function Neural Networks with Randomly Generated Hidden Nodes for Knowledge based Pattern Recognition
지식 기반 패턴 인식용 무작위 생성 히든 노드를 갖춘 소프트맥스 방사형 기반 함수 신경망 설계
In this research, we proposed a new classification method which is an expanded version of the generic radial basis function neural networks. In the proposed classification method, the expanded version of RBFNNs consists of two main parts. And the first part is the soft-max function to help us handle the output distribution as a multinomial distribution. Secondly, radial basis functions, which can be considered as hidden nodes in the conventional neural networks, are defined not by using a clustering algorithm but in a random manner. When the soft-max function is applied to the networks, iterative learning algorithms such as nonlinear least square estimation and error back-propagation algorithm should be used to value the coefficient of the networks. In order to validate the proposed classification method, several machine learning data sets are used.
한국어
본 연구에서는 일반 방사형 기저 함수 신경망을 확장한 새로운 분류 방법을 제안한다. 제안된 분류 방법에서 RBFNN의 확장 버전은 두 가지 주요 부분으로 구성된다. 첫 번째 부분은 출력 분포를 다항 분포로 처리하는 데 도 움이 되는 소프트 맥스 함수이다. 둘째, 기존 신경망에서 히든노드로 간주될 수 있는 방사형 기반 함수를 클러스터링 알고리즘을 사용하지 않고 무작위로 정의한다. 소프트맥스 함수를 네트워크에 적용할 경우 비선형 최소자승추정, 오 류역전파 알고리즘 등의 반복학습 알고리즘을 사용하여 네트워크의 계수 값을 구해야 한다. 제안된 분류 방법을 검 증하기 위해 여러 가지 기계 학습 데이터 세트가 사용된다.
목차
Abstract 요약 Ⅰ. Introduction Ⅱ. Radial Basis Function Neural Networks Related Work 2.1 Generic RBFNNs 2.2 RBFNNs with RBFs defined by Fuzzy Clustering Algorithm Ⅲ. Soft-Max Radial Basis Function Neural Networks with Random Generated Hidden 3.1 Nonlinear Least Square Estimation 3.2 Back-Propagation Ⅳ. Experimental Studies 4.1 Toy Example 4.2 Machine Learning Data Ⅴ. Conclusions REFERENCES
키워드
방사형 기반함수 신경망소프트맥스 함수비선형 최소자승추정무작위 생성 히든노드지식 기반 패턴 인식Radial Basis Function Neural NetworksSoft-Max functionNonlinear Least Square EstimationRandomly Generated Hidden NodesKnowledge based pattern recognition
저자
Dong-Yoon Lee [ 이동윤 | Professor, Department of Electrical and Electronic Engineering, Joongbu University ]
Corresponding Author
Ever since next generation convergence technology became one of the most important industries in the nation, computing professionals have encountered a growing number of challenges. Along with scholars and colleagues in related fields, they have gathered in avariety of forums and meetings over the last few decades to share their knowledge, experiences and the outcome of their research. These exchanges have led to the founding of the International Next-generation Convergence technology (INCA) on December 1, 2015. INCA was registered as an incorporated association under the Ministry of Information and Communications. The main purpose of the organization is to improve our society by achieving the highest capability possible in next generation convergence technology.
간행물
간행물명
차세대융합기술학회논문지 [The Journal of Next-generation Convergence Technology Association]