Algorithm 1 Reinforcement learning with evolution strategies.
1: Given
2:     Parent NN with weight matrix $W^{\left(i\right)}$	$i=1,2$
3:     number of children m
4:     learning_rate $\eta$
5: Start
6:     for iteration in a predefined range do
7:         for h in range m do
8:            $Chil{d}^{\left(h\right)}=$ Parent NN + random noise	($W^{\left(i\right)\left(h\right)}=W^{\left(i\right)}+noise$)
9:            Evaluate $Chil{d}^{\left(h\right)}\to Rewar{d}^{\left(h\right)}$
10:         Calculate $Mean\_reward$
11:         $Gai{n}^{\left(h\right)}=Rewar{d}^{\left(h\right)}-Mean\_reward$	$h=1,\dots ,m$
12:         Parent NN → Parent NN + $\eta \times {\sum }_{h=1}^{m}Gai{n}^{\left(h\right)}\times Chil{d}^{\left(h\right)}$	$W^{\left(i\right)}=\eta \times \sum Gai{n}^{\left(h\right)}\times W^{\left(i\right)\left(h\right)}$
13:         Evaluate Parent NN
14: End
Return the highest performing Parent NN