Evolving an Artificial Neural network without using backpropagation

2013 3 27
1 个回答
回答已采纳
2 次查看（30 天）

0 个投票

I am looking for a simple code to evolve an artificial neural network without using backpropagation. I want to evolve both the weights and the activation function. Thanking you in advance

2 个评论
显示 无 隐藏 无

Greg Heath
Greg Heath 2013-3-27
Please explain what you mean by evolving an activation function.
It sounds like a big waste of time.
Greg
Tinashe
Tinashe 2013-3-27
I mean, usually one would evolve weights and bias. But in my case, I want to compare 5 activation functions.

请先登录，再进行评论。

请先登录，再回答此问题。

 采纳的回答

Greg Heath
Greg Heath 2013-3-28

0 个投票

If you wish to compare training functions, use a straight forward backprop e.g., fitnet or patternnet.
Thank you for accepting my answer
Greg

更多回答（0 个）

类别

在 帮助中心 和 File Exchange 中查找有关 Deep Learning Toolbox 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by