Archived

This topic is now archived and is closed to further replies.

NN Perceptron questions

This topic is 5148 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I have been reading about NNs, and I mostly understand them, conceptually. I''m looking to implement one, but I have a few things I don''t understand. 1) The learning constant...how is it determined and what exactly does it control? 2) I don''t quite understand modifying weights. Specifically, I was looking at the article, "An Introduction To Neural Networks", at the OR neuron. The weights are shown in a sort of graph form, and I don''t really get it. I also don''t get how they''re modifying the threshold value (the weight which is not related to an input).

Share this post


Link to post
Share on other sites
If your talking about ''modified'' in terms of during training, then the bias is changed just like every other weight during backprop.

You can think of the bias as a weight with a constant input (1). 1*bias=bias. Thefore when looking at how to change it, you just treat it like any other weight on the network. The formula does the rest.

Share this post


Link to post
Share on other sites