Technology
eCortex has two principle technologies: the Emergent software package for neural network simulation, and the Leabra algorithms that run in that software.
Emergent is a GUI interface for creating neural network simulations in a transparent way. It has been our primary research tool, and under very active development since its creation in 2006. It incorporates a vast array of analysis and network creation tools, and focuses on making the function of our networks transparent. The graphical interface makes understanding and changing network function a visual and intuitive process. While Emergent is open-source and freely available for academic purposes, eCortex holds an exclusive license for using it commercially. Emergent is extensively documented here.
The other critical aspect of our technology is the Leabra algorithm. The name stands for Local, Error-driven and Associative, Biologically Realistic Algorithm. This set of algorithms represents the best cumulative theory of brain function that currently exists. Since it uses a biologically-realistic variant of the back propagation learning algorithm, Leabra represents an important bridge between brain function and the increasingly impressive power of deep back-propagation networks. This theory is explained in detail in Dr. O'Reilly's online textbook, and more briefly in this article.
Emergent is a GUI interface for creating neural network simulations in a transparent way. It has been our primary research tool, and under very active development since its creation in 2006. It incorporates a vast array of analysis and network creation tools, and focuses on making the function of our networks transparent. The graphical interface makes understanding and changing network function a visual and intuitive process. While Emergent is open-source and freely available for academic purposes, eCortex holds an exclusive license for using it commercially. Emergent is extensively documented here.
The other critical aspect of our technology is the Leabra algorithm. The name stands for Local, Error-driven and Associative, Biologically Realistic Algorithm. This set of algorithms represents the best cumulative theory of brain function that currently exists. Since it uses a biologically-realistic variant of the back propagation learning algorithm, Leabra represents an important bridge between brain function and the increasingly impressive power of deep back-propagation networks. This theory is explained in detail in Dr. O'Reilly's online textbook, and more briefly in this article.
