info
Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Cloning Hidden Nodes
8 views
Skip to first unread message
Stephen Wolstenholme
Jul 25, 2022, 11:32:06 AM7/25/22
to
I have recently been comparing the learning speed of a very large
network with and without hidden node cloning. The problem is comparing
my health symptoms with my diet. The inputs are the food and
supplement items and the outputs are the symptoms. There are 970 daily
examples to learn. The network has 729 input nodes, 9 output nodes and
a single layer of 26 hidden nodes. With that number of hidden nodes
the network learns 970 daily examples to a target error of 0.01 in 51
minutes. If hidden node cloning is allowed the learning time is less
than a minute. When the target error is reached 3 more hidden nodes
have been added. The difference on smaller networks is not
significant. I have decided to set cloning by default.
Steve
--
Neural Network Software for Windows http://www.npsnn.com
network with and without hidden node cloning. The problem is comparing
my health symptoms with my diet. The inputs are the food and
supplement items and the outputs are the symptoms. There are 970 daily
examples to learn. The network has 729 input nodes, 9 output nodes and
a single layer of 26 hidden nodes. With that number of hidden nodes
the network learns 970 daily examples to a target error of 0.01 in 51
minutes. If hidden node cloning is allowed the learning time is less
than a minute. When the target error is reached 3 more hidden nodes
have been added. The difference on smaller networks is not
significant. I have decided to set cloning by default.
Steve
--
Neural Network Software for Windows http://www.npsnn.com
0 new messages