Ensemble Network Architecture for Deep Reinforcement Learning

Algorithm 1

The temporal and target values ensemble algorithm.

Initialize action-value network with random weights

Initialize the target neural network buffer

For episode 1, do

For , do

With probability select a random action , otherwise

Execute action in environment and observe reward

and next state , and store transition ) in

Sample random minibatch of transition ) from

set

Ensemble -learner

set

set

set

Set

Every steps reset

End for

End for