News Bits

by Luke Muehlhauser on July 10, 2011 in News

From Twitter:


Previous post:

Next post:

{ 5 comments… read them below or add one }

MarkD July 10, 2011 at 8:38 pm

The Norvig v. Chomsky discussion takes me back to the 90s when IBM’s Candide MT system first demonstrated improvements on Transfer-based systems. I was asked the same question and found myself puzzled as to exactly what linguistic theory really tells us when subjected to the same kinds of deep analysis. The existence of different roles for language symbols and the recursive nature of the generative grammar doesn’t really resolve any existence questions concerning language other than possibly emphasizing the notion that given a finite set of symbols it is possible to generate an infinite number of utterances which means that human creativity is unbounded (in practice, the long tail rules).

Perhaps the most interesting machine learning work that impinges on language concerns attempts to simulate language learning and acquisition. Work on statistical language learning that specifically tests against (measurable) child acquisition phenomena includes the application of Minimum Description Length (MDL) and Maximum Entropy approaches for coping with the “poverty of stimulus” that remains a vexing challenge for understanding language learning. In some ways Chomsky’s own Minimalist programme is tracking that work, amusingly enough.


mojo.rhythm July 10, 2011 at 9:12 pm

Zizek is a very charming guy; he also makes some killer points. Assange is also brilliant but he is not a natural stage guy like Zizek.


Reginald Selkirk July 11, 2011 at 9:14 am

Out-Of-Body Experiences Linked to Neural Instability and Biases in Body Representation

This result is consistent with the hypothesis that OBEs are an illusion.


cl July 11, 2011 at 12:31 pm

How often we find exactly what we’re looking for!


Baal July 11, 2011 at 9:59 pm

Then it’s probably a good sign that I failed absolutely in having any kind of OBE despite years of trying in my youth. I was just too neurologically stable.


Leave a Comment