goessl / neuralnetwork

Very basic neural network implementation.

NeuralNetwork

Very basic neural network implementation to understand & try out machine learning. In this repository you can find two neural network classes, one without & one with biases. Both work mostly the same and support the same methods so they will referred to as the neural network. I recommend using the bias version for better results.

The neural network class is only used for construction, forward & backward propagation! To train the network sebig3000/Optimization is needed!

Usage

Construct a new neural network by calling its constructor and specifying its parameters. These include the number of inputs, the number of nodes in every layer and the activation functions of every layer & their derivatives. Common activation functions & their derivatives are already implemented in ActivationFunction. Now you can already forward datasets through the network & do back propagation.

Input and output is stored in matricies in which every row represents an dataset and every column gets feed into an node (e.g. element[1][2] is a parameter of a dataset with index [1] and gets feed into the node with index [2])

To train the network you create a new NeuralNetworkOptimizer and give it your neural network. hen call whatever optimization algorithm you like with your input & output sets and let the optimizer train the network. (Adam is recommended)

Getting Started

This repository depends on sebig3000/Matrix! So make sure you have already installed it. Then you can simply download this repository and add it to your project as a new package! Done!

Contributors

The two people who inspired me to try making my own machine learning project are Brandon Rohrer and Stephen Welch. Both make awesome YouTube videos that explain how machine learning works.

Stephen Welch:

• YouTube: https://www.youtube.com/user/Taylorns34
• Homepage: http://www.welchlabs.com/
• GitHub: https://github.com/stephencwelch

Brandon Rohrer:

• YouTube https://www.youtube.com/channel/UCsBKTrp45lTfHa_p49I2AEQ
• Blog: https://brohrer.github.io/blog.html
• GitHub: https://github.com/brohrer

TODO

• Fix dependencies
• Remove optimization wrappers
• Argument validation
• Documentation

License (MIT)

MIT License

Copyright (c) 2019 Sebastian Gössl

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

neuralnetwork's Issues

It does not seem to work

It does not seem to work, it generates random output values every time you try to use...
I try to learn this simple XOR boolean operation and does not work :

// Training Data
// 0.0, 0.0, 1.0, 1.0
// 0.0, 1.0, 0.0, 1.0
XOR
// Expected results
// 0.0, 1.0, 1.0, 0.0