For anyone interested, I threw together a higher level interface for Sebastian's ArcadeLearningEnvironment.jl:
https://github.com/tbreloff/AtariAlgos.jl It's unregistered, and will probably stay that way unless people beg me to do it. On Thu, Dec 10, 2015 at 11:06 PM, Tom Breloff <[email protected]> wrote: > What great timing! I just posted this ( > https://groups.google.com/forum/?hl=en#!topic/julia-users/Up2STYjtwdE), > and I was going to start work on it myself since no one responded. > Thanks! I'll check it out. > > > On Thursday, December 10, 2015 at 6:07:30 PM UTC-5, Sebastian Nowozin > wrote: >> >> >> Hi, >> >> the arcade learning environment (ALE), >> http://www.arcadelearningenvironment.org/, is a modified Atari 2600 >> emulator that provides additional in-game information (such as scores and >> game-over states) for around 50 classic Atari 2600 games. >> It is a popular environment for testing artificial intelligence agents >> and for reinforcement learning methods. >> >> The ArcadeLearningEnvironment.jl package is a thin wrapper around ALE and >> exposes all functionality. >> It is based on the same underlying C library used by the Python wrapper. >> >> It is available here: >> https://github.com/nowozin/ArcadeLearningEnvironment.jl >> I only tested the package on Linux (where it downloads and builds the ALE >> libraries), but on other architectures you need to manually provide the >> libale_c.dll or libale_c.so file. >> >> To use the package you need to obtain Atari 2600 ROM files supported by >> ALE, a list of these is available at the ALE homepage. >> >> Any feedback is welcome. >> >> Sebastian >> >
