Stamatis N, Parthimos D, Griffith T M
Department of Radiology, University of Wales College of Medicine, Cardiff, U.K.
IEEE Trans Biomed Eng. 1999 Dec;46(12):1441-53. doi: 10.1109/10.804572.
A multilayer perceptron (MLP) network architecture has been formulated in which two adaptive parameters, the scaling and translation of the postsynaptic function at each node, are allowed to adjust iteratively by gradient-descent. The algorithm has been employed to predict experimental cardiovascular time series, following systematic reconstruction of the strange attractor of the training signal. Comparison with a standard MLP employing identical numbers of nodes and weight learning rates demonstrates that the adaptive approach provides an efficient modification of the MLP that permits faster learning. Thus, for an equivalent number of training epochs there was improved accuracy and generalization for both one- and k-step ahead prediction. The applicability of the methodology is demonstrated for a set of monotonic postsynaptic functions (sigmoidal, upper bounded, and nonbounded). The approach is computationally inexpensive as the increase in the parameter space of the network compared to a standard MLP is small.
已经构建了一种多层感知器（MLP）网络架构，其中允许两个自适应参数，即每个节点处突触后函数的缩放和平移，通过梯度下降进行迭代调整。在对训练信号的奇异吸引子进行系统重构之后，该算法已被用于预测实验性心血管时间序列。与采用相同节点数量和权重学习率的标准MLP进行比较表明，自适应方法对MLP进行了有效的改进，从而允许更快的学习。因此，对于相同数量的训练轮次，一步和k步超前预测的准确性和泛化能力都得到了提高。该方法对于一组单调突触后函数（Sigmoid函数、上界函数和无界函数）的适用性得到了证明。由于与标准MLP相比，网络参数空间的增加很小，因此该方法在计算上成本较低。
