In an interview, Numenta founder Jeff Hawkins says he expects his vision of brain-inspired artificial intelligence (AI) to win out over other approaches. He predicts by the end of the 2020s, universal, memory-based algorithms that work on numerous problems will dominate, and they will be founded on time-based patterns and serve as online learning paradigms. The algorithms will operate according to what Hawkins calls hierarchical temporal memory, which is designed to mirror neocortical function. "Our basic approach is adhering to neuroscience principles so that we will get the properties that brains have," Hawkins says.
He also notes Numenta's technology can support computer-vision tasks, creating a cortical-like vision system that incorporates both sensory motor inference and high order inference. Hawkins says his AI approach is distinct from deep learning, which involves the solving of spatial pattern-recognition problems with no consideration of time-based patterns or prediction or anomaly detection. Still, he acknowledges the need for his approach and other AI approaches to ultimately intersect.
Hawkins describes the technology Numenta is developing as "a foundation for the next 60, 70, 100 years of computing. It's not replacing computing, but it's as big as computing."
View Full Article
Abstracts Copyright © 2014 Information Inc., Bethesda, Maryland, USA