This repository contains a variant of implementing a standard MLP with two layers for educational purposes. Relative simplicity of this particular implementation was aimed to show the constructions upon which some more advanced models are based, how forward and backward passes are computed and etc. In future versions an arbitrary number of layers and neurons per layer will be allowed.
git clone https://github.com/iworeushankaonce/mlp.git- MLP class with arbitrary layers
- Implement important activation functions
- 'Leaky ReLU';
- 'ELU';
- 'linear';
- 'softmax';
- 'sigmoid'.
- Implement important loss functions
- 'binary_crossentropy';
- 'sparse_categorical_crossentropy';
- 'categorical_crossentropy';
- 'MSE';
- 'MAE'.
- Allow usage of GD, Batch GD, Nesterov Accelerated Gradient, Momentum
- Provide 'try-it-yourself' Notebook with famous XOR example
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
Please make sure to update tests as appropriate.