Does the train function always reinitialize a neural network

1 次查看（过去 30 天）

Marshall Worth 2017-2-15

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/325206-does-the-train-function-always-reinitialize-a-neural-network

评论： Marshall Worth 2017-2-15

So I see that calling the train function will initialize a neural network, but will it reinitialize the same network (ie different set of random weights and biases) every time it is called? I have a network where new data is constantly being produced and would like to train/update the network with the new targets and inputs that I store over time. Should I be using the adapt function? I would prefer the train function as it seems that I would have to run some kind of error checking/comparison of inputs and targets every time I use adapt to know when my training is "done".

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Marshall Worth 2017-2-15

Seems that my answer lies within this previously asked question...

https://www.mathworks.com/matlabcentral/answers/92991-is-there-a-difference-between-incremental-training-using-adapt-and-train-in-neural-network-toolbox-6

请先登录，再进行评论。

请先登录，再回答此问题。