simple-neural-network
simple-neural-network copied to clipboard
mattm
A simple Python script showing how the backpropagation algorithm works.
Hi, could you expand the example on how to calculate error derivatives to adjust weights when having more than one hidden layer? Thanks!
A very clear example except there is no update to bias. Highly suggest to update that. As said by many others, it is very useful.
Fixed a currently existing issue with the repo, the bias values now get updated while training
This just links to a 404 page,
when calculate the output delta value, it should be==> pd_errors_wrt_output_neuron_total_net_input[h] = self.output_layers.neurons[h].calculate_pd_error_wrt_output(training_outputs[h])*self.output_layers.neurons[h].calculate_pd_total_net_input_wrt_input() or change the function calculate_pd_error_wrt_output() ??
I have attempted to add support multiple layers to the network demo. I did my best to maintain a similar style to that used by the original author.
As this require python 3 and based on others, it is suggested to add this. Also there is no update to bias which some other implementation and course may want...
Hi Matt, Would you be willing to add utf-8 support and running from terminal to your great introductory script, so that one could download and run it as-is (some machines...
Hi, do you have a pdf version of a-step-by-step-backpropagation-example? I can't load many diagrams in the page. Thank you. My email is [email protected]
