[Read me obsolete] This project has been made to make implementation of neural network in C# easy.
Here's the step to implement a neural network (at the time only feed-forward neural network) with this library :
[1] To create a new neural network, you just need to call its constructor and feed it the parameters :
- Training data (must be array of inputs (itself an array of double) double[][])
- Expected Results Data (must be array of outputs (itself an array of double) double[][])
- Number of hidden layers
- Number of neurons per layer
- Activation function (find some on the MathTools static class that comes with the project)
- Activation function for outputs (find some on the MathTools static class that comes with the project)
- Loss function (find some on the MathTools static class that comes with the project)
- Training rate (must be double)
[2] Call Train() method of the neural network created in 1) to train the neural network.
[3] Call Predict() method of the neural network to predict, it takes an input double[] and return an output double[] of the prediction.