Stability Analysis of Limit Cycles on Continuous-time Cyclic Connection Neural Networks

연속시간 모델 순환결합형 신경회로망에서의 리미트사이클의 안정성 해석

  • 박철영 (대구대학교 전자공학부)
  • Published : 2006.04.01


An intuitive understanding of the dynamic pattern generation in asymmetric networks may be considered an essential component in developing models for the dynamic information processing. It has been reported that the neural network with cyclic connections generates multiple limit cycles. The dynamics of discrete time network with cyclic connections has been investigated intensively. However, the dynamics of a cyclic connection neural network in continuous-time has not been well-known due to the considerable complexity involved in its calculation. In this paper, the dynamic behavior of a continuous-time cyclic connection neural network, in which each neuron is connected only to its nearest neurons with binary synaptic weights of ${\pm}1$, has been investigated. Furthermore, the dynamics and stability of the network have been analyzed using a piece-wise linear approximation.

신경회로망을 동적 정보처리에 응용하기 위해서는 비대칭 결합 신경회로망에서 생성하는 동적 상태천이에 관한 직관적 이해가 필요하다. 자기결합을 갖고 결합하중치가 비대칭인 순환결합형 신경회로망은 복수 개의 리미트사이클을 기억할 수 있다는 것이 알려져 있다. 현재까지 이산시간 모델의 네트워크에 대한 상태천이 해석은 상세하게 이루어져 왔다. 그러나 연속시간모델에 대한 해석은 네트워크 규모의 증가에 따른 급격한 계산량의 증가 때문에 그다지 활발하게 연구가 이루어지지 않고 있다. 본 논문에서는 각 뉴런이 최근접 뉴런에만 이진화된 결합하중 +1 및 -1로 연결된 연속시간모델 순환결합형 신경회로망의 동적인 상태천이 특성을 해석하여 이산시간 모델에서 기억 가능한 리미트사이클과의 차이점을 분석하였다. 또한 뉴런의 활성화 함수가 완전선형인 경우와 구분선형 근사인 네트워크에 대한 리미트사이클의 안정성을 해석하였다.


  1. D. J. Amit, H. Gutfreund and H. Sompolinsky, 'Storing infinite number of patterns in spin-glass model of neural networks,' Phys, Rev.. Lett. 55, pp. 1530-1533, 1985
  2. J. H. Li, A. N. Michel and W. Porod, 'Qualitative analysis and synthesis of a class of neural network,' IEEE Trans. CAS, vol. 35, no. 8, pp. 976-986, 1988
  3. Y. Mori, P. Davis and S. Nara, 'Pattern retrieval in an asymmetric neural network with embedded limit cycles', Journal of Phys., vol. A22, pp. L525-532, 1989
  4. K. Nowara and T. Saito, 'Guaranteed storing of limit cycles into a discrete-time asynchronous neural network,' IEICE Trans. on Fundamentals, vol. E75-A, pp. 1579-1582, 1992
  5. Y. Hayashi, 'Oscillatory neural network and learning of continuously transformed patterns,' Neural Networks, vol. 7, pp. 219-231, 1994
  6. G. Setti, P. Thiran and C. Serpico, 'An approach to information propagation in 1- D Cellular Neural networks-part Global propagation II,' IEEE Trans. on CAS I : Fundamental theory and applications, Vol 45, No .8, pp. 790-811, 1998
  7. H. Sompolinsky and I. Kanter, 'Temporal association in asymmetric neural networks,' Phys. Rev. Lett., vol. 57, pp. 2861-2864, 1986
  8. K. Aihara, M. Takanabe, and M. Toyoda, 'Chaotic neural networks,' Phys. Lett. A, vol. 44, pp. 333-340, 1990
  9. J. F. Kolen and S. C. Kremer, 'A field guide to dynamical recurrent networks,' IEEE Press, New York, 2001
  10. Y. Hayakawa, A. Marumoto, and Y Sawada, 'Effect of the noise in the performance of a neural network model for optimization problems,' Phys. Rev. E, vol. 51, pp. 2693-2696, 1995
  11. K. Nakajima and Y. Hayakawa, 'Correct Reaction Neural Network,' Neural Networks, vol. 6, pp. 217-222, 1993
  12. C. Y. Park, Y. Hayakawa, K. Nakajima and Y. Sawada, 'Limit cycles of one-dimensional neural networks with the cyclic connection matrix,' IEICE Trans. on Fundamentals, vol. E79-A, no. 6, pp, 752-757, 1996
  13. C. Y. Park, and K. Nakajima, 'Asymptotic Analysis of Cyclic Transitions in the Discrete-Time Neural Networks with Antisymmetric and Circular Interconnection Weights,' IEICE Trans. on Fundamentals, Vol. E87-A, No.6, pp. 1487-1490, 2004
  14. H. Ninomiya, A. Kamo, T. Yoneyama, and H. Asai, 'A Fast Algorithm for Spatiotemporal Pattern Analysis of Neural Networks with Multivalued Logic,' IEICE Trans. on Fundamentals, vol. E81A, no. 9, pp. 1847-1852, 1998