Neural Network written from scratch in C++ with unit testing (using google test). Currently supports:
- Creating and training a feedforward neural network with SGD using the MSE as the loss
Future roadmap:
- Add functionality for easily handling and training on data 🔘
- Add more activation functions
- Add more loss functions
- Optimize code, including getting it to work on GPU's
- Different optimizers (eg. SGD with momentum, Adam, etc.)
- Different layer types (eg. convolutional)
- Examples that use the library