AleControl 0.10.0.122-beta1

The AleControl gives Windows C# developers easy access to the Arcade-Learning-Environment (ALE)[1]. ALE is a modification of the Atari-2600 Emulator[2] from the Stella Team that provides access to numerous ATARI games (such as Pong, Space Invaders, etc) for Reinforcement Learning. The games run actions provided by the user and produce their overall game visualizations and game state.

For more information on ALE, please see https://github.com/mgbellemare/Arcade-Learning-Environment

For more information on Stella, please see https://github.com/stella-emu/stella

The AleControl uses the 'atari_win64' source tree which is a fork off the ALE Github tree that has been modified to run as a Windows 64-bit DLL and is licensed under the GNU license.

The 'atari_win64' project uses the Simple DirectMedia Layer (SDL for short) which is a cross-platform library designed to make it easy to write multi-media software such as games and emulators.

The Simple DirectMedia Layer library source code is available from: http://www.libsdl.org, and the SDL library is distributed under the terms of the GNU LGPL License.

The AleControl, written by SignalPop LLC, is a Windows 64-bit COM control that gives any OLE Automation enabled language (C#, Visual Basic, etc.) easy access to the ALE envrionment via OLE Automation and is licensed under the Apache 2.0 license. An extensive list of ATARI game ROM files is provided by OpenAI on Github at openai/atari-py/atari_roms and are distributed under the GNU GPL License. For the full project, visit us on GitHub at AleControl.

When used in combination with MyCaffe (A complete C# re-write of CAFFE[3]) the AleControl can be used to solve Reinforcement Learning related problems via the MyCaffeTrainerRL control. You can also use Nuget to get MyCaffe.

The SignalPop AI Designer provides a development environment allows you to quickly pull all of these parts together to visually design MyCaffe based models that are both compatible with native CAFFE and support Reinforcement Learning for the Arcade-Learning-Environment.

Supported Development Environments:

• Visual Studio 2017
• Visual Studio 2015

References

[1] The Arcade Learning Environment: An Evaluation Platform for General Agents by Marc G. Bellemare, Yavar Naddaf, Joel Veness and Michael Bowling, 2012-2013. Source code available on GitHub at mgbellemare/Arcade-Learning-Environment

[2] Stella - A multi-platform Atari 2600 VCS emulator by Bradford W. Mott, Stephen Anthony and The Stella Team, 1995-2018. Source code available on GitHub at stella-emu/stella

[3] CAFFE: Convolutional Architecture for Fast Feature Embedding by Yangqing Jai, Evan Shelhamer, Jeff Donahue, Sergey Karayev, Jonathan Long, Ross Girshick, Sergio Guadarrama, and Trevor Darrell, 2014. Source code available on Github at BVLC/caffe