Hmm... The biggest deficiency in related work, that I am aware of, has to do with an inability to handle context. Symbols are ambiguous and more-or-less meaningless without context.
Meanwhile, reports on brain activity when a person is reacting to a joke suggests that the brain uses a sort of hypothesis/confirm/reject mechanism for handling linguistic recognition - perhaps somewhat related to simulated annealing (but with something approximating a scheduled cutoff)? Then again, given how clueless and awful people can be, I'm not entirely sure that machine learning by itself is necessarily going to always be a fruitful activity - I imagine that any such effort will need support from (presumably mediocre but not entirely inadequate) fallback mechanisms. Still, interesting to hear about. Thanks, -- Raul On Mon, Nov 9, 2015 at 1:29 PM, Skip Cave <[email protected]> wrote: > In reference to the recent discussions on Machine Learning and AI, I > ran across this interesting article.. > > Shortened URL: > http://goo.gl/6daIhc > > Full URL > http://techcrunch.com/2015/11/09/google-open-sources-the-machine-learning-tech-behind-google-photos-search-smart-reply-and-more/ > > Skip > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
