It sounds like Google is looking at voice recognition much the way I suggested optical character recognition should be done on mobile devices the other day: use the device as a sensor that uploads raw data over a high-speed network to a server that performs the heavy lifting of pattern recognition. The slow improvement of battery technology has put a crimp on delivering the kind of processor power required for pattern recognition on mobile devices, but it doesn't seem to have hurt the growth of wireless network bandwith. I don't expect a return to thin client mobile applications like some Web 2.0 evangelists do, but I do see a big future for the web for service delivery to mobile applications and even mobile software platforms.
According to reader Brian, Jeff Hawkins' Numenta neuroscience technology project has already produced "an extremely powerful machine vision system that could do OCR and *much* more (objects, shapes, faces, etc)." Hawkins apparently spoke repeatedly about applications for machine vision in mobile devices at a recent Wall Street Journal technology conference. With this kind of progress being shown and Numenta being headed up by two of the founders of the original Palm company, you have to wonder if the secret "Third Business" of Palm that Hawkins keeps hinting at doesn't have to do with the use of Numenta's Hierarchical Temporal Memory systems to make mobile devices smarter. Whenever he drops hints about what he's working on in his capacity at Palm he always encourages us to consider the implications of massive storage and always-connected broadband wireless for mobile devices. It seems to me one of the biggest implications of these facts is access to powerful crunching of complex data that servers can provide--just what you need for pattern recognition. Are Hawkin's two jobs at Numenta and Palm more closely related than we first supposed?
Posted by cervezas at 07:08:00. Filed under: Mobile Technology
Comments
Add Comment