Algorithm 1: GraphSAGE embedding generation (i.e., forward propagation) algorithm
Input	Graph $G(V,\epsilon )$; input features $\left\{x_v,\forall v\in V\right\}$; depth K; weight matrices $W^k, \forall k\in \left\{1,...K\right\}$; non-linearity $\sigma$; differentiable aggregator functions ${AGGREGATE}_k, \forall k \in \left\{1,...K\right\}$; neighbourhood function $N:v\to 2^V$
Output	Vector representation $z_v$ for all $v\in V$
1	$h_v^0\leftarrow x_v,\forall v\in V$;
2	For $k=1...K$ do
3		For $v\in V$ do
4			$h_{N_{\left(v\right)}}^k\leftarrow {AGGREGATE}_k\left(\left\{h_u^{k-1},\forall u\in N_{\left(v\right)}\right\}\right)$,
5			$h_v^k\leftarrow \sigma \left(W^k·CONCAT\left(h_u^{k-1},h_{N_{\left(v\right)}}^k\right)\right)$,
6		end
7		$h_v^k\leftarrow h_v^k/{‖h_v^k‖}_2,\forall v\in V$,
8	end
9	$z_v\leftarrow h_v^k,,\forall v\in V$,