February 12, 2012 · 10:00 AM
This location is shown only to members
The Brain Emulating Cognition and Control Architecture (BECCA) is a set of algorithms for reinforcement learning and actual selection designed for allowing robots to learn in a similar way that creatures with brains learn and demonstrate general intelligence. It's currently implemented in MATLAB, with a port to Python in the works, and has demonstrated some very basic ability to allow simple robotics hardware to learn basic behaviors given a defined goal or task.
This event will include some demos of capabilities, a review of the published literature, and an introduction to hacking the code base and creating your own Agents and Tasks.
The main program will be a presentation of BECCA's architecture in a way that is not necessarily specific to a particular programming language. Example code will primarily be MATLAB, however.
Anyone interested in staying for an 'overtime' hands-on tutorial with MATLAB should bring a laptop with an installation of GNU Octave (the open-source implementation of the MATLAB language):
Windows & Mac installers:
BECCA Homepage: http://www.sandia.gov/~brrohre/becca.html