This repository includes homemade C++ source code of deep learing
- neurons is a static library including some basic neuron concepts.
Following algorithms and functions are included in this module
- vector and matrix calculations
- activation functions
- cost or error functions
- convolutional functions
- pooling functions
- forward and backward propagation
- Back Propagation Through Time
- basic building block of RNN (RNN_unit)
- One dimensional GMM EM algorithm
- dataset is a module with a general dataset interface for neural network training.
Supports of following dataset are included in this module
- MNIST for feedforward neural networks
- CIFAR_10 for feedforward neural networks
- Media Review for recurrent neural networks (RNN)
- dnn is a multi-layer fully connected neural network program that can be trained.
- cnn is a homemade multi-layer convolutional neural network program that can be trained.
- rnn has recently implemented a simple RNN module. LSTM is still under construction.
- test includes some basic test cases of neural calculations.