Reinforcement Learning (Actor Critic)
#18
by odunbar - opened
No description provided.
Adds the results from Reinforcement learning methods:
- Two methods DDPG and TD3, and
- Two timescales 1xwindow and 0.1xwindow for updates.
- Uses gamma = 0.5, and some other sensible RL network hyperparameters for now. some sweeping has been undertaken!
MISC
- Add the % failure in suitability table when more than 20% failed
- Added the new taxonomy classification gradient(kalman) and gradient(score) etc.
odunbar changed pull request status to open
odunbar changed pull request status to merged