I want to train multiple feedforward neural network simultaneously with various combination of inputs and after that I want to add their individual output....Is it poosible in matlab...then please hel me ....

Question

S Saha 2013-1-12

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/58627-i-want-to-train-multiple-feedforward-neural-network-simultaneously-with-various-combination-of-input

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

José-Luis 2013-1-12

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/58627-i-want-to-train-multiple-feedforward-neural-network-simultaneously-with-various-combination-of-input#answer_70917

Yes, it is possible.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

Answer 2

Greg Heath 2013-1-13

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/58627-i-want-to-train-multiple-feedforward-neural-network-simultaneously-with-various-combination-of-input#answer_70947

A neural net ensemble is created by combining the outputs of multiple nets. Although you can combine posterior probability estimates for classification/patternnrecognition, ensembles are typically used for regression/curvefitting.

A neural net committee is created by combining the {0,1} votes of multiple classifiers.

SIMULTANEOUSLY training multiple nets can be accomplished by deleting connections in a single net so that there are no mixing of signals among hidden and output layers.

This requires both intializing hidden-to-hidden and hidden-to-output weights to zero and setting their learning rates to zero. Zeroing input-to-hidden weights is optional.

The simplest example I can think of is a 3-output classifier for the 3-class/4-dimensional iris_dataset (load iris_dataset; whos ).

The hidden nodes are separated into 3 categories. The hidden nodes in one category are only connected to the output node for one class and are trained to separate that class from the other two.

This topoloogy is only interesting if the combination of inputs connected to each category is different.

Finally, the outputs of those 3 independent classifiers can be combined into an ensemble or committee.

Hope this helps.

Thank you for formally acceptingmy answer.

Greg