Skip to content
master
Go to file
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

README.md

Learning Multimodal Transition Dynamics for Model-Based Reinforcement Learning

Code for reproducing key results in the paper Learning Multimodal Transition Dynamics for Model-Based Reinforcement Learning by Thomas M. Moerland, Joost Broekens and Catholijn M. Jonker.

Prerequisites

  1. Install recent versions of:
  • Python 3
  • Tensorflow
  • Numpy (e.g. pip install numpy)
  • Matplotlib
  1. Clone this repository:
git clone https://github.com/tmoer/multimodal_varinf.git

Syntax

Example:

python3 vae_main.py --logdir <logdir> --hpconfig network=1,n_rep=10,var_type='discrete',K=3,N=3,verbose=False
python3 vae_grid.py --logdir <logdir> --hpconfig network=1,n_epochs=75000,n_rep=5,var_type='continuous',z_size=8,n_flow=0,artificial_data=False,use_target_net=True,test_on_policy=True,verbose=False

For default hyper-parameters, look at the get_hps() function in the vae_grid.py and vae_main.py scripts.

Reproducing Paper Results

Run:

bash paper_toy.sh (Sec 4.1)
bash paper_grid.sh (Sec 4.2)
bash paper_grid_rl.sh (Sec 4.2)

Citation

@proceedings{moerland2017learning,
	author = "Moerland, Thomas M. and Broekens, Joost and Jonker, Catholijn M.",
	note = "arXiv preprint arXiv:1705.00470",
	journal = "Scaling Up Reinforcement Learning (SURL) Workshop @ European Machine Learning Conference (ECML)",
	title = "{Learning Multimodal Transition Dynamics for Model-Based Reinforcement Learning}",
	year = "2017"
}

About

Code for paper "Learning Multimodal Transition Dynamics for Model-Based Reinforcement Learning".

Resources

License

Releases

No releases published

Packages

No packages published
You can’t perform that action at this time.