Among hese method , one of the simplest is the deepest gradient descent algorithm , but it has one disadvantage that the solutions are oscillatory and slowly convergent . a friction term is ntroduced to avoid oscillatory solutions and raise convergence rate . the third major contribution of this dissertation is study on 3 - d diffuse object surface optical automatic measurement 其次，结合要处理的具体问题，介绍数值优化及几种常用的无约束最优化计算方法，以最优梯度法为基础，讨论其算法的优缺点，通过引进阻尼系数，改进了算法的收敛速度，从而提高迭代精度。
2 to overcome the shortcoming of the interference capture caused by the least square cma ( ls - cma ) under severe environment , we provide the preprocessing ls - cma which combine the advantage of the stochastic gradient descent cma and the ls - cma . the simulation results confirm its effectiveness and feasibility . with the preprocessing controlled by the stochastic gradient descent cma , the ls - cma rejects the interference capture and converges to the user signal with weak power 2针对最小二乘恒模算法在强干扰环境下产生干扰捕获的缺点，结合随机梯度恒模算法和最小二乘恒模算法的特点，提出了预处理最小二乘恒模算法，计算机仿真结果证明，经过随机梯度恒模算法控制的预处理过程，最小二乘恒模算法克服了干扰捕获的缺点，收敛于弱功率用户信号，并取得了较好的信干噪比。
A new hybrid learning algorithm is proposed to train the fuzzy neural network based on tsk fuzzy model . firstly , fuzzy c - means algorithm is applied to initialize the parameters of the fuzzy neural network . secondly , the parameters of the premise part of the fuzzy rule are learned by the gradient descent algorithm 提出了一阶tsk模糊神经网络的混合学习算法，算法由三部分组成：基于模糊聚类的网络初始化；基于梯度下降的规则前件的学习算法；基于部分最小二乘的规则后件的学习算法。
A soft sensor modeling algorithm based on improved fuzzy neural network is presented . the normalized average output membership functions are defined as fuzzy basis functions for defuzzification calculation . in order to improve the property of convergence , some parameters of the fuzzy neural network are trained by levenberg - marquardt algorithm , and the others are trained by gradient descent algorithm 提出了一种改进的模糊神经网络软测量建模方法，采用规则化的平均输出隶属度函数作为模糊基函数进行反模糊化运算；在训练网络时，部分参数采用levenberg - marquardt算法来训练，另一部分采用一阶梯度下降法。
Using the indices of trajectory follow error and steering busyness in evaluation of the steering stability , a quadratic form performance index function for the neuron learning was established and the tuning of the connection weight values of the single neuron controller was realized using the algorithm of gradient descent 利用操纵稳定性评价中的轨迹跟随误差和方向盘忙碌程度的评价指标，建立了神经元学习的二次型性能指标函数，并采用梯度下降算法实现了单神经元控制器的滚接权值的调整。
