Perception and Cortical Learning
The human cortex is by far the best learning system in existence. It learns more efficiently, by several orders of magnitude, than any artificial system to date. We seek to understand the computational principles by which it works, in ways that will allow us to reproduce that success.
Our primary project exploring those principles is our Leabra Vision model, which rapidly learns to categorize objects in images. While this technology has many useful applications as a computer vision, it is primarily a means for us to validate the Leabra theory of cortical function. One main line of research for eCortex is on how the brain learns to perform its impressive feats. Visual object recognition is our testbed for those theories. We choose this focus because empirical data such as single-cell recording is most abundant in that domain.
Select related publications:
Our primary project exploring those principles is our Leabra Vision model, which rapidly learns to categorize objects in images. While this technology has many useful applications as a computer vision, it is primarily a means for us to validate the Leabra theory of cortical function. One main line of research for eCortex is on how the brain learns to perform its impressive feats. Visual object recognition is our testbed for those theories. We choose this focus because empirical data such as single-cell recording is most abundant in that domain.
Select related publications:
- The Leabra Cognitive Architecture: How to Play 20 Principles with Nature and Win! O'Reilly, R.C., Hazy, T.E. & Herd, S.A. (in press). S. Chipman (Ed) Oxford Handbook of Cognitive Science,Oxford: Oxford University Press.
- Recurrent Processing during Object Recognition O'Reilly, R.C., Wyatte, D., Herd, S., Mingus, B. & Jilk, D.J. (2013). Frontiers in Psychology, 4, 124.