This is a PyTorch implementation of Adversarial Inverse Reinforcement Learning(AIRL)[1] and GAIL[2] based on DDPG[2] for mine master thesis. The implementation took the code references of [3]-[10]. Also thank you for Prof. Dr. Heinz Köppl[11] and supervisor[12] giving me this chance to implement my ideas as graduation work.
You can install Python liblaries using pip3 install -r requirements.txt.
train forward model:
python3 train_rl.py
train inverse model:
python3 train_imitation.py
run online server: extract rswebsite.zip, install maven dependencies in pom.xml then run main function in RecServer. Open http://localhost:6016/ in browser.
Run deployed model via TensorFlow Serving(in windows), docker download link: https://www.docker.com/. Open docker and then:
docker run -t --rm -p 8501:8501 -v ".\rswebsite\src\main\resources\webroot\modeldata\rl:/models/recmodel" -e MODEL_NAME=recmodel tensorflow/serving
[1] Fu, Justin, Katie Luo, and Sergey Levine. "Learning robust rewards with adversarial inverse reinforcement learning." arXiv preprint arXiv:1710.11248 (2017).
[2] Timothy P. Lillicrap et.al: "Continuous control with deep reinforcement learning".
[3] machine learning of mofan
[4] Discriminator-Actor-Critic: Addressing Sample Inefficiency and Reward Bias in Adversarial Imitation Learning
[5] gail-airl-ppo.pytorch project by Toshiki Watanabe
[6] ILSwiss is an Easy-to-run Imitation Learning (IL, or Learning from Demonstration, LfD) and also Reinforcement Learning (RL) framework (template) in PyTorch.
[7] Simple implementations of various popular Deep Reinforcement Learning algorithms using TensorFlow2
[8] A Deep Learning Recommender System
[9] Deep Reinforcement Learning based Recommender System by Young_Painter_L,kyunghoon-jung and pasus
[10] practice on movielens using pytorch
[11] Prof. Dr. Heinz Köppl
[12] Matthias Schultheis
[13] Haarnoja, Tuomas, et al. "Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor." arXiv preprint arXiv:1801.01290 (2018).\