API Reference - Models
Recurrent Deep Reinforcement Learning
-
Note that all of these recurrent models require RecurrentNeuralNetworkCell or GatedRecurrentUnitCell containers. It is recommended to use the former since it uses less computational resources than the latter.
-
Currently, these recurrent models have no documentation. Fortunately, you can still refer to the non-recurrent versions of these models.
-
Additionally, they cannot work with DataPredict’s QuickSetups for deep reinforcement learning. You’ll have to use the classic setup to use the recurrent models.
| Model | Alternate Names | Use Cases |
|---|---|---|
| RecurrentDeepQLearning | Recurrent Deep Q Network | Self-Learning Fighting AIs, Self-Learning Parkouring AIs, Self-Driving Cars |
| RecurrentDeepDoubleQLearningV1 | Recurrent Double Deep Q Network (2010) | Same As Recurrent Deep Q-Learning |
| RecurrentDeepDoubleQLearningV2 | Recurrent Double Deep Q Network (2015) | Same As Recurrent Deep Q-Learning |
| RecurrentDeepClippedDoubleQLearning | Recurrent Clipped Double Deep Q Network | Same As Recurrent Deep Q-Learning |
| RecurrentDeepStateActionRewardStateAction | Recurrent Deep SARSA | Same As Recurrent Deep Q-Learning |
| RecurrentDeepDoubleStateActionRewardStateActionV1 | Recurrent Double Deep SARSA | Same As Recurrent Deep Q-Learning |
| RecurrentDeepDoubleStateActionRewardStateActionV2 | Recurrent Double Deep SARSA | Same As Recurrent Deep Q-Learning |
| RecurrentDeepExpectedStateActionRewardStateAction | Recurrent Deep Expected SARSA | Same As Recurrent Deep Q-Learning |
| RecurrentDeepDoubleExpectedStateActionRewardStateActionV1 | Recurrent Double Deep Expected SARSA | Same As Recurrent Deep Q-Learning |
| RecurrentDeepDoubleExpectedStateActionRewardStateActionV2 | Recurrent Double Deep Expected SARSA | Same As Recurrent Deep Q-Learning |
| RecurrentMonteCarloControl | None | Same As Recurrent Deep Q-Learning |
| RecurrentOffPolicyMonteCarloControl | None | Same As Recurrent Deep Q-Learning |
| RecurrentVanillaPolicyGradient | Recurrent VPG | Same As Recurrent Deep Q-Learning |
| RecurrentREINFORCE | None | Same As Recurrent Deep Q-Learning |
| RecurrentActorCritic | Recurrent AC | Same As Recurrent Deep Q-Learning |
| RecurrentAdvantageActorCritic | RecurrentA2C | Same As Recurrent Deep Q-Learning |
| RecurrentSoftActorCritic | Recurrent SAC | Same As Recurrent Deep Q-Learning |
| RecurrentProximalPolicyOptimization | Recurrent PPO | Same As Recurrent Deep Q-Learning |
| RecurrentProximalPolicyOptimizationClip | RecurrentPPO-Clip | Same As Recurrent Deep Q-Learning |
| RecurrentDeepDeterministicPolicyGradient | Recurrent DDPG | Same As Recurrent Deep Q-Learning |
| RecurrentTwinDelayedDeepDeterministicPolicyGradient | Recurrent TD3 | Same As Recurrent Deep Q-Learning |
BaseModels
RecurrentReinforcementLearningBaseModel
RecurrentReinforcementLearningActorCriticBaseModel