GoNeuralNetworks

Still Major Work in Progress

Golang Neural Networks Stuffs

Networks

• MnistDataFF (DONE): This is just a simple feed forward neural network doing it's thing on a mnist data set. Just making sure the feed forward neural network works before I try anything crazy. (dataset is in mnist_dataset folder zipped up)
• HandwritingFF (IN PROGRESS): Me trying to get this bad boy to be able to learn all handwritten letters from a more. dataset

Packages

• core: The core logic and data structures for the neural networks (training, predictions, ...)
• data: The logic and data structures for loading and processing the data for the network (loading images from disk, turning image into matricies, ...)
• util: Super general utilities (printing *mat.Dense, getting random number generator, ...)
• main: The actual code that sets up, trains, and tests the networks using the other packages

Cool Stuff

• Images are loaded from disk and pre-processed in parrall using goroutines
• Implemented from scratch using GoNum
• That's about it. This is still just me making an artificial neural network in golang and that's about it

Dependencies

• GoNum: https://github.com/gonum/gonum
• Image Resizing: https://github.com/nfnt/resize

Refrences

• https://sausheong.github.io/posts/how-to-build-a-simple-artificial-neural-network-with-go/
• Shit load of stuff from https://stackoverflow.com/ (as any programmer does)