DeepLume

A deep learning library mockup implemented from scratch in python, with numpy being the only library used. The code implements the forward pass and backward error propagation methods for:

2 Types of Layers:

1. Dense Layer
2. Dropout Layer

3 Activation Functions:

1. Linear
2. ReLU
3. Softmax

4 Loss Functions:

1. Categorical Crossentropy
2. Binary Crossentropy
3. Mean Squared
4. Mean Absolute

4 Optimizers:

1. Stochastic Gradient Descent
2. Adaptive Gradient Descent
3. Root Mean Squared Propagation
4. Adam