Real-world uses

Created for this library

1. 1.

   A robotics research team uses a critic network to estimate value while the actor network selects actions in a manipulation policy.

2. 2.

   A trading research team uses an actor-critic setup where the critic learns the expected return of each action in a market state.

3. 3.

   An RL platform team adds a learned critic to an existing policy-gradient agent to reduce variance during training of a logistics dispatcher.