Skip to content

Neural network builder/trainer utilizing no machine learning packages. Structure is customizable. Output nodes and input nodes must all be numeric, however. https://machinelearningmastery.com/implement-backpropagation-algorithm-scratch-python/ as reference for methodology, but all code is original.

Notifications You must be signed in to change notification settings

Dillon214/DilloNN_first_try

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 

Repository files navigation

DilloNN_1

Neural network builder/trainer. I wrote this simply to explore the algorithm more thouroughly, and gain a better apreciation for what's at play behind the popular modules. Unfinished, but currently functional. Currently cannot have multiple output nodes, and the output node it does have must be a linear activation (This goes against the guide I used, but I wanted the network to predict numbers, not categories). Used https://machinelearningmastery.com/implement-backpropagation-algorithm-scratch-python/ as reference, but all code is original.

Version 2 can have multiple output nodes, and an error in calculating bias was fixed.

Version 3 handles all elements in arrays rather than individually. Thhis boosts overall speed by a factor of how ever many elements are being handled. Output nodes are still exclusively linear.

About

Neural network builder/trainer utilizing no machine learning packages. Structure is customizable. Output nodes and input nodes must all be numeric, however. https://machinelearningmastery.com/implement-backpropagation-algorithm-scratch-python/ as reference for methodology, but all code is original.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages