cavis/rl4j/README.md

87 lines
2.9 KiB
Markdown

# RL4J: Reinforcement Learning for Java
RL4J is a reinforcement learning framework integrated with deeplearning4j and released under an Apache 2.0 open-source license. By contributing code to this repository, you agree to make your contribution available under an Apache 2.0 license.
* DQN (Deep Q Learning with double DQN)
* Async RL (A3C, Async NStepQlearning)
Both for Low-Dimensional (array of info) and high-dimensional (pixels) input.
![DOOM](doom.gif)
![Cartpole](cartpole.gif)
Here is a useful blog post I wrote to introduce you to reinforcement learning, DQN and Async RL:
[Blog post](https://rubenfiszel.github.io/posts/rl4j/2016-08-24-Reinforcement-Learning-and-DQN.html)
[Examples](https://github.com/eclipse/deeplearning4j-examples/tree/master/rl4j-examples)
[Cartpole example](https://github.com/eclipse/deeplearning4j-examples/blob/master/rl4j-examples/src/main/java/org/deeplearning4j/examples/rl4j/Cartpole.java)
# Disclaimer
This is a tech preview and distributed as is.
Comments are welcome on our gitter channel:
[gitter](https://gitter.im/deeplearning4j/deeplearning4j)
# Quickstart
** INSTALL rl4j-api before installing all (see below)!**
* mvn install -pl rl4j-api
* [if you want rl4j-gym too] Download and mvn install: [gym-java-client](https://github.com/eclipse/deeplearning4j/tree/master/gym-java-client)
* mvn install
# Visualisation
[webapp-rl4j](https://github.com/rubenfiszel/webapp-rl4j)
# Quicktry cartpole:
* Install [gym-http-api](https://github.com/openai/gym-http-api).
* launch http api server.
* run with this [main](https://github.com/rubenfiszel/rl4j-examples/blob/master/src/main/java/org/deeplearning4j/rl4j/Cartpole.java)
# Doom
Doom is not ready yet but you can make it work if you feel adventurous with some additional steps:
* You will need vizdoom, compile the native lib and move it into the root of your project in a folder
* export MAVEN_OPTS=-Djava.library.path=THEFOLDEROFTHELIB
* mvn compile exec:java -Dexec.mainClass="YOURMAINCLASS"
# Malmo (Minecraft)
![Malmo](malmo.gif)
* Download and unzip Malmo from [here](https://github.com/Microsoft/malmo/releases)
* export MALMO_HOME=YOURMALMO_FOLDER
* export MALMO_XSD_PATH=$MALMO_HOME/Schemas
* launch malmo per [instructions](https://github.com/Microsoft/malmo#launching-minecraft-with-our-mod)
* run with this [main](https://github.com/eclipse/deeplearning4j-examples/blob/master/rl4j-examples/src/main/java/org/deeplearning4j/examples/rl4j/MalmoPixels.java)
# WIP
* Documentation
* Serialization/Deserialization (load save)
* Compression of pixels in order to store 1M state in a reasonnable amount of memory
* Async learning: A3C and nstep learning (requires some missing features from dl4j (calc and apply gradients)).
# Author
[Ruben Fiszel](http://rubenfiszel.github.io/)
# Proposed contribution area:
* Continuous control
* Policy Gradient
* Update gym-java-client when gym-http-api gets compatible with pixels environments to play with Pong, Doom, etc ..