how did you define your action space?

carpedm20 / deep-rl-tensorflow

TensorFlow implementation of Deep Reinforcement Learning papers

MIT License

1.6k stars 396 forks source link

how did you define your action space? #33

Open OsgoodWu opened 6 years ago

OsgoodWu commented 6 years ago

After I read your code carefully, I cannot figure out how you define your action space. For example , how many actions you define and how to represent each action? Waiting for your answers.

Sincerely

danielkaifeng commented 6 years ago

I am also waiting for answers on this. In deep Q learning, the target Qt is calculate as target_q_t = (1. - terminal) * self.discount_r * max_q_t_plus_1 + reward, which means the action space is one dimension? If it is two dimension or above, numpy matric can't conduct multiply and add method as this in the code.