NeuralCPP: A Neural Network Library in C++

NeuralCPP is a Neural Network Library written in C++. It offers a simple API to create and train neural networks.

Table of Contents

1. Installation
2. Example
3. Hierarchical Structure
4. Documentation
5. Libraries Used
6. See Also
7. License

Installation

To add NeuralCPP to your project, follow these steps:

1. Add NeuralCPP as a submodule to your project:

git submodule add -b main https://github.com/B-Manitas/NeuralCPP.git
git submodule update --init --recursive

2. Include the NeuralCPP/include/NeuralCPP.hpp file in your project.

3. Compile your project with the following flags:

-std=c++11 -fopenmp

Example

#include "include/NeuralCPP.hpp"

int main()
{
    // Create the dataset
    cmatrix<float> X, y;
    NeuralCPP::create_dataset(X, y, 100000, 2, 2);

    // Create the neural network model (2 hidden layers with 32 neurons each)
    NeuralLayers nn({32, 32});

    // Train the model
    nn.fit(X, y, 10000, .01, true);

    // Predict the output
    cmatrix<cbool> y_pred = nn.predict(X);

    return 0;
}

Hierarchical Structure

NeuralCPP is structured as follows:

Class	Description
include
NeuralCPP.hpp	Includes all the other headers.
NeuralLoss.hpp	Defines the loss functions.
NeuralActivation.hpp	Defines the activation functions.
NeuralPerceptron.hpp	The Perceptron model.
src
	This folder contains the implementation of the library.

Documentation

For detailed information on how to use CMatrix, consult the documentation.

Libraries Used

• CMatrix: A C++ library for matrix operations. (Required for compile CMatrix)
• OpenMP: An API for parallel programming. (Required for compile CMatrix)
• Doxygen: A documentation generator.
• Matplot++: A C++ Graphics Library for Data Visualization.

See Also

• CMatrix: A C++ library for matrix operations.
• CDataFrame: A C++ DataFrame library for Data Science and Machine Learning projects.

License

This project is licensed under the MIT License, ensuring its free and open availability to the community.