Definition: Feedforward NN

FNN is an architecture which is defined by a DAG (Directed Acyclic Graph)

DAG: $G \sim (V, A)$ meaning: the graph contains the vertices and arches.

where: $V = {v_{1}, v_{2}, \dots, v_{n}}$ (vertices).

and, $A = {a_{1}, a_{2}, \dots, a_{m}}$ (arches).

with: $a_{k} \sim (v_{i}, v_{j})$ (arch from $v_{i}$ to $v_{j}$ ).

$a_{k} \sim (v_{i}, v_{j})$ is permitted only if $i ≺ j$ ( $i$ precedes $j$ ) in the relationships.

Backpropagation

The Backpropagation algorithm works only for FNN. Even if they have more than 1 hidden layer (MLN - Multi Layered Neural Networks)

MLN are FNN