Research: From Palmtops to Brain Cells

From Palmtops to Brain Cells
Economist Technology Quarterly (03/08) Vol. 386, No. 8570, P. 31

Palm Pilot creator and Numenta founder Jeff Hawkins aspires to make computers work in a manner that more closely resembles the human brain through his theory of hierarchical temporary memory, which posits that the brain processes information using hierarchically organized pattern-recognition "nodes." Frequently-observed patterns are identified and learned over time by nodes at each hierarchical level, and when an established pattern triggers a node, it sends a signal to the next level up in the hierarchy. As multiple signals ascend the hierarchy, nodes at higher levels learn to recognize and anticipate more sophisticated patterns, and predictions are passed down the hierarchy so that disparities between predicted and observed patterns can be identified. The Numenta Platform for Intelligent Computing is an expression of Hawkins' model in software, and Hawkins hopes the free toolkit will be applied toward the development of software that functions more like the human brain. Such software could find use in a diverse array of fields that includes robotics, video games, data analysis, and computer vision. New York University computer scientist Yann LeCun says enthusiasm for the creation of intelligent machines has waned among the machine-learning community over the past decade, and Hawkins' work is rekindling interest in the concept among younger researchers. Although he admires Hawkins' intuition, University of Toronto professor Geoffrey Hinton thinks Hawkins is underestimating the inherent difficulty of creating algorithms capable of mimicking intelligence.
Click Here to View Full Article