【摘 要】
：
使用新的分类子网,改进了Huang得到的一个关于两个隐层前馈型神经网络的学习能力的结果,证明了具有2(2N)(1/2)+2个隐层节点的神经网络可以学会N个不同的样本.同时,新的讨论
【机 构】
：
上海理工大学管理学院 上海200093
论文部分内容阅读
使用新的分类子网,改进了Huang得到的一个关于两个隐层前馈型神经网络的学习能力的结果,证明了具有2(2N)(1/2)+2个隐层节点的神经网络可以学会N个不同的样本.同时,新的讨论方法使得结果对广泛的一类活化函数都适用,而不只限于sigmoid活化函数.
Using the new classification subnet, Huang improved the learning ability of two hidden layer feedforward neural networks, and proved that the neural network with 2 (2N) (1/2) +2 hidden nodes N different samples can be learned, and at the same time the new method of discussion makes the results suitable for a wide range of activation functions, not just the sigmoid activation function.
其他文献
Three different punches are designed for the hydropiercing experiments and finite element simulations are conducted by finite element program ABAQUS-3D to inves
With the continuous development of cloud manufacturing technology, in order to solve more complex manufacturing problem and conduct large-scale networked manufa
The performance of an annular jet pump ( AJP ) is determined by its area ratio A ( ratio of cross sectional area of throat and annular nozzle) and flow rate rat
The uniform mathematical model of distortion signals in power grid has been setup with the theory of Wiener-G Functional. Firstly, the Matlab simulation models
In order to consider the influence of the continuous rotary motor electro-hydraulic servo system parameters change on its performance, the design method of back
A design method of an intelligent temperature control system based on single-chip microcomputer is presented in this paper. The intelligent temperature control
The hot deformation behavior of Mg2B2O5w/Al composite was investigated by compression test at different temperatures and different strain rates. The constitutiv
人类对潜在有毒污染物的接触及其健康风险是当今环境科学与医学共同关心的热点问题.而污染土壤中有毒污染物在土壤-植物系统迁移与转化是模型计算结果表明,未污染条件下的土
One of the major technical challenges in using carbon dioxide ( CO2 ) as part of the cushion gas of the underground gas storage reservoir ( UGSR) is the mixture
本文认为要切实推进我国产业结构调整 ,必须突破需求、技术、人力、体制和利益五大障碍
This paper argues that we must break through the five major obstacles of deman