연상 메모리 기능을 수행하는 셀룰라 신경망의 설계 방법론

A Design Methodology for CNN-based Associative Memories

  • 박연묵 (한국과학기술원 기계공학과) ;
  • 김혜연 (고려대학교 영상정보처리학과) ;
  • 박주영 (고려대학교 제어계측공학과) ;
  • 이성환 (고려대학교 컴퓨터학과)
  • 발행 : 2000.05.15

초록

본 논문에서는 연상 메모리 기능을 수행하는 셀룰라 신경망(Cellular Neural Network)의 설계를 위한 새로운 방법론을 제안한다. 먼저, 셀룰라 신경망 모델의 기본적 특성들을 소개한 후, 최적 성능을 가지고 이진 원형 패턴들을 저장할 수 있는 셀룰라 신경망 모델의 설계 방법을 제약 조건이 가해진 최적화 문제로 공식화한다. 다음으로 이 문제의 제약 조건을 선형 행렬 부등식(Linear Matrix Inequalities)을 포함하는 부등식의 형태로 변환시킬 수 있음을 관찰한다. 마지막으로 셀룰라 신경망 최적 설계 문제를 내부점 방법(interior point method)에 의해 효율적으로 풀릴 수 있는 일반화된 고유값 문제(Genaralized EigenValue Problem)로 변환한다. 본 논문에서 제시하는 셀룰라 신경망 설계 방법론은 공간 변형 형판 셀룰라 신경망과 공간 불변 형판 셀룰라 신경망 설계에 모두 적용될 수 있다. 설계 예제를 통해 제안된 방법의 유효성을 검증한다.

In this paper, we consider the problem of realizing associative memories via cellular neural network(CNN). After introducing qualitative properties of the CNN model, we formulate the synthesis of CNN that can store given binary vectors with optimal performance as a constrained optimization problem. Next, we observe that this problem's constraints can be transformed into simple inequalities involving linear matrix inequalities(LMIs). Finally, we reformulate the synthesis problem as a generalized eigenvalue problem(GEVP), which can be efficiently solved by recently developed interior point methods. Proposed method can be applied to both space varying template CNNs and space-invariant template CNNs. The validity of the proposed approach is illustrated by design examples.

키워드

참고문헌

  1. Roska, T., and Vandewalle, J. (Eds.), Cellular Neural Networks, John Wiley & Sons, New York, 1995
  2. Chua, L. O., and Yang, L., Cellular neural networks: Theory and Applications, IEEE Transactions on Circuits and Systems 35, 1257-1290, 1988 https://doi.org/10.1109/31.7600
  3. Gahinet, P., Nemirovskii, A., Laub, A. J., and Chilali, M., LMI Control Toolbox, The Mathworks Inc., Natick, MA, 1995
  4. Liu, D., and Michel, A. N., Cellular neural networks for associative memories, IEEE Transactions on Circuits and Systems II 40, 119-121, 1993 https://doi.org/10.1109/82.219843
  5. Liu, D., and Michel, A. N., Robustness analysis and design of a class of neural networks with sparse interconnecting structure, Neurocomputing 12, 59-76, 1996 https://doi.org/10.1016/0925-2312(95)00040-2
  6. Michel, A. N. and Farrell, J. A., Associative memories via artificial neural networks, IEEE Control Systems Magazine 10, 6-17, 1990 https://doi.org/10.1109/37.55118
  7. Vandenberghe, L., and Balakrishnan, V., Algorithms and software for LMI problems in control, IEEE Control Systems Magazine 17, 89-95, 1997 https://doi.org/10.1109/37.621480
  8. Nesterov, Y., and Nemirovskii, A., Interior Point Polynomial Methods in Convex Programming, SIAM, Philadelphia, PA, 1994
  9. Wu, C. W., and Chua, L. O., More rigorous proof of complete stability of cellular neural networks, IEEE Transactions on Circuits and Systems-I 44, 370-371, 1997 https://doi.org/10.1109/81.563629
  10. Liu, D., and Lu, Z., A new synthesis approach for feedback neural networks based on the perceptron training algorithm, IEEE Transactions on Neural Networks 8, 1468-1482, 1997 https://doi.org/10.1109/72.641469
  11. Boyd, S., ElGhaoui, L., Feron, E., and Balakrishnan, V., Linear Matrix Inequalities in Systems and Control Theory, SIAM, Philadelphia, PA, 1994
  12. Chan, H. Y., and Zak, S. H., Real-time synthesis of sparsely interconnected neural associative memories, Neural Networks 11, 749-759, 1998 https://doi.org/10.1016/S0893-6080(98)00015-X
  13. Liu, D., Cloning template design of cellular neural networks for associative memories, IEEE Transactions on Circuits and Systems-I 44, 646-650, 1997 https://doi.org/10.1109/81.596948
  14. Lu, Z., and Liu, D., A new synthesis procedure for a class of cellular neural networks with space-invariant cloning template, IEEE Transactions on Circuits and Systems-II, to appear https://doi.org/10.1109/82.746682