Deep Feed Forward Networks or Multilayer perceptrons are function approximator that approximate some function $f()$. A feedforward network defines a mapping between $y$ and $f(x, \theta)$ and learns a value of the parameter $\theta$
In the linked example a feed forward neural network is constructed, trained and evaluated on a dataset
This is a demo project that uses state-of-the art computer vision models like resnet to classify waste products in order to raise awareness about wish recycling