There are many, but one plausible rule is based on the idea that weight and threshold modification should be determined by a fraction of (t - y).
有很多，不过有一条似乎合理的规则是基于这样一种思想，即权系数和阈值的调整应该由分式(t- y)确定。
The relation between threshold, weight and matching degree is also discussed. In order to make the querying friendlier, the method to avoid returning null set and corpora is also presented.
研究了权重、阈值和匹配度之间的关系，提出了避免查询结果为空集或全集的方法，使得查询更加人性化。
This paper develops a novel threshold signature scheme based on probability weight.
提出了一种新颖的基于概率权重的门限签名方案。
This RPE algorithm was adapted to the simple recurrent network by making improvement and complementarity, and the weight and the threshold of the network can be adjusted at the same time.
对RPE算法进行了改进和补充，使之适用于简单递归网，用来对网络的权值和阈值进行调整。
What an interpreter needs to do for whole interpretation is only to provide some parameters, such as time window, threshold, weight coefficients etc.
解释员只需提供时窗、门坎值、加权系数等几个参数，就可自动地实现全过程的解释。
Selected standard samples are trained, the stable weight values and threshold value of the system are obtained.
用选定的标准样本进行训练，得到稳定的权值和阈值。
Moreover, weight threshold is employed to increase the validity of pre-sampling.
同时利用权值阈值估计增加预采样的有效性。
The nonlinear error of PSD was modified by choosing the centre of RBF, ascertaining the number of neural cell of the neural network and adjusting the weight and the threshold of each hiberarchy.
该方法通过选择径向基函数中心、确定神经网络隐层神经元的数目和调整每一层的权值和阈值，对由于PSD非线性产生的误差进行修正。
RBF neural network adopts the off-line training and the on-line adaptation of weight and threshold value. In order to speed up the convergence, the grads descent method with inertia item was used.
RBF神经网络采用离线学习在线修正权值和阈值，为加快收敛速度，应用带惯性项的梯度下降法。
RBF neural network adopts the off-line training and the on-line adaptation of weight and threshold value. In order to speed up the convergence, the grads descent method with inertia item was used.
RBF神经网络采用离线学习在线修正权值和阈值，为加快收敛速度，应用带惯性项的梯度下降法。
应用推荐