DOI QR코드

DOI QR Code

Design of Digit Recognition System Realized with the Aid of Fuzzy RBFNNs and Incremental-PCA

퍼지 RBFNNs와 증분형 주성분 분석법으로 실현된 숫자 인식 시스템의 설계

  • Kim, Bong-Youn (Department of Electrical Engineering, The University of Suwon) ;
  • Oh, Sung-Kwun (Department of Electrical Engineering, The University of Suwon) ;
  • Kim, Jin-Yul (Department of Electronic Engineering, The University of Suwon)
  • Received : 2016.01.19
  • Accepted : 2016.02.22
  • Published : 2016.02.25

Abstract

In this study, we introduce a design of Fuzzy RBFNNs-based digit recognition system using the incremental-PCA in order to recognize the handwritten digits. The Principal Component Analysis (PCA) is a widely-adopted dimensional reduction algorithm, but it needs high computing overhead for feature extraction in case of using high dimensional images or a large amount of training data. To alleviate such problem, the incremental-PCA is proposed for the computationally efficient processing as well as the incremental learning of high dimensional data in the feature extraction stage. The architecture of Fuzzy Radial Basis Function Neural Networks (RBFNN) consists of three functional modules such as condition, conclusion, and inference part. In the condition part, the input space is partitioned with the use of fuzzy clustering realized by means of the Fuzzy C-Means (FCM) algorithm. Also, it is used instead of gaussian function to consider the characteristic of input data. In the conclusion part, connection weights are used as the extended diverse types in polynomial expression such as constant, linear, quadratic and modified quadratic. Experimental results conducted on the benchmarking MNIST handwritten digit database demonstrate the effectiveness and efficiency of the proposed digit recognition system when compared with other studies.

본 연구에서는 퍼지 RBFNNs과 증분형 주성분 분석법으로 실현된 숫자인식 시스템의 설계를 소개한다. 주성분 분석법은 차원축소를 위해 사용되는 알고리즘으로 학습데이터의 차원 수가 고차원이거나 데이터의 양이 많을 때 특징 추출을 위한 많은 계산 시간을 필요로 한다. 따라서 고차원 데이터의 효율적인 차원축소와 점진적인 학습을 위해 증분형 주성분분석법을 적용하는 방법을 제안한다. 방사형 기저함수 신경회로망의 구조는 조건부, 결론부, 추론부의 3가지 기능적 모듈로서 구분이 가능하다. 조건부에서는 FCM 클러스터링 알고리즘의 도움으로 실현된 퍼지 클러스터링의 사용으로 입력 공간을 분할한다. 또한 가우시안 함수 대신 FCM(Fuzzy C-Means)클러스터링 알고리즘의 멤버쉽 값을 사용함으로써 입력 데이터의 특성을 좀 더 잘 반영할 수 있도록 개선하였으며, 결론부에서 연결가중치는 상수항에서 일차식과 이차식, 그리고 변형된 이차식과 같은 다항식의 형태로 확장하여 사용한다. 실험 결과는 공인 숫자 데이터인 MNIST 필기체 숫자 데이터를 사용하여 제안된 숫자 인식 시스템의 효율성을 다른 연구와의 비교를 통해 입증한다.

Keywords

References

  1. S. K. Oh, W. Pedrycz, B. J. Park, "Polynomial-based Radial Basis Function Neural Networks realized with the Aid of Particle Swarm Optimization," Fuzzy Sets and Systems, vol. 163, pp. 54-77, 2011. https://doi.org/10.1016/j.fss.2010.08.007
  2. S. K. Oh, W. D. Kim, and W. Pedrycz, "Fuzzy Radial Basis Function Neural Networks with information granulation and its parallel genetic optimization," Fuzzy Sets and Systems, vol. 237, pp. 96-117, February 2014. https://doi.org/10.1016/j.fss.2013.08.011
  3. H. H lee, X. Cui, H. R. Kim, S. W. Ma, J. H. Lee and H. I. Kim, "Robust Object Tracking in Mobile Robots using Object Features and On-line Learning based Particle Filter," Journal of Institute of Control, vol. 18, pp. 562-570, 2012.
  4. Byoung-Joo Kim, "Feature Extraction on High Dimensional Data Using Incremental PCA," KIICS Integrated Conference, vol. 8, no. 7, pp. 1475-1479, 2004.
  5. D. Zhang and Z. H. Zhou, "$(2D)^2PCA$ : Two-directional two-dimensional PCA for efficient face representation and recognition," Neurocomputing, vol. 69, pp. 224-231, 2005. https://doi.org/10.1016/j.neucom.2005.06.004
  6. Y. LeCun, L. Bottou, Y. Bengio and P. Haffner, "Gradient-Based Learning Applied to Document Recognition," IEEE, vol. 86, pp. 2278-2324, 1998. https://doi.org/10.1109/5.726791
  7. S. C. Lee, S, K. Oh and H. K. Kim, "Design of PCA-based pRBFNNs Pattern Classifier for Digit Recognition," Korea Institute of intelligent systems, vol. 25, no. 4, pp. 355-360, 2015. https://doi.org/10.5391/JKIIS.2015.25.4.355
  8. S. H. Lee, S. Y. Choi, S. Y. Lee and Y. S. Kim, "Licence Plate Recognition Using Improved IAFC Fuzzy Neural Network," Korea Institute of intelligent systems, vol. 19, no. 1, pp. 6-12, 2009. https://doi.org/10.5391/JKIIS.2009.19.1.006
  9. S. S. Yoon, H. S. Chung, K. O. Yi, Y. B. Lee and S. H. Lee, "A Study on Human Recognition Experiments with Handwritten Digit for Machine Recognition of Handwritten Digit," Korea Institute of intelligent systems, vol. 18, no. 3, pp. 373-380, 2008. https://doi.org/10.5391/JKIIS.2008.18.3.373
  10. In-Cheol Kim, "Word Separation in Handwritten Legal Amounts on Bank Check by Measuring Gap Distance Between Connected Components," Korea Institute of intelligent systems, vol. 14, no. 1, pp. 57-62, 2004. https://doi.org/10.5391/JKIIS.2004.14.1.057
  11. W. Y. Choi, S. K. Oh and H. K. Kim, "Design of RBFNN-Based Pattern Classifier for the Classification of Precipitation/Non-Precipitation," Korea Institute of intelligent systems, vol. 24, no. 6, pp. 586-591, 2014. https://doi.org/10.5391/JKIIS.2014.24.6.586
  12. Yann LeCun, Corinna Cortes and Christopher J.C. Burges, "The MNIST DATABASE of handwritten digits," Available:http://yann.lecun.com/exdb/mnist

Cited by

  1. Design of Pattern Classifier for Electrical and Electronic Waste Plastic Devices Using LIBS Spectrometer vol.26, pp.6, 2016, https://doi.org/10.5391/JKIIS.2016.26.6.477