A highly customizable lightweight library on Neural Network.

Unlike its counterparts (i.e., Lasagne, Keras), Parable builds a "shallow" (or centralized) network structure. Most Neural Network libraries store weights and manage weights inside various Layer classes (such as Conv2DLayer() or DenseLayer()). Parable however, stores and manages weights and biases inside the Network class, and distribute those into various layers when expression are being evaluated.