用连续回归神经网络求解泛函极值问题
A CONTINUOUS TIME RECURRENT NEURAL NETWORK BASED METHOD TO SOLVE FUNCTIONAL MINIMIZATION PROBLEM
-
摘要: 针对信息科学和控制理论中经常涉及的一类泛函极值问题,提出基于连续回归神经网络的求解方法。推导了求解泛函的连续BPTT算法,进而对该算法进行改进,得出一种在线学习算法,为并行实现打下了基础.Abstract: In this paper, the continuous time recurrent neural network is proposed to solve the functional minimization problem, which is often involved in estimation and control. At first, the continuous time BPTT algorithm corresponding to the problem is presented. Then,an on-line algorithm based on the amendments of the BPTT algorithm is discussed. This on-line algorithm paves the way for parallel realization.
-
叶庆凯,郑应干编著.变分法及其应用.国防工业出版社,1991.[2]Funahashi K,Nakamura Y.Approximation of dynamical systems by continuous time recurrent neural networks[J].Neural Network.1993,6:801-806[3]Rumelhart D E,Hinton G E,Williams R J.Learning internal representations by error propagation.in Parallel Distributed Processing.Rumelhart,D.E.,McClelland,J.L.Eds.,Cambridge,MA:M.I.T Press,1986. [4]Werbos P J.Backpropagation through time:What it does and how to do it,Proc.of IEEE,1990,78(10):1550-1560.[4]Pearlmutter B A.Learning state space trajectories in recurrent neural network,IEEE Proc.IJCNN,1989,2:365-372.[5]Sato M.A Learning algorithm to teach spatiotemporal patterns to recurrent neural networks[J].Biological Cybernetics.1990,62:259-263[6]Williams R J,Zipser D.A Learning algorithm for continually running fully recurrent neural networks.Neural Computation.1989,1(2):270-280.[7]Baldi P.Gradient learning algorithm overview:A general dynamical systems perspective.IEEE Trans.on Neural Networks.1995,6(1):182-195.
计量
- 文章访问数: 2313
- HTML全文浏览量: 74
- PDF下载量: 552
- 被引次数: 0