SUTTON Richard
<p>May 25, 2011 10:31 PM
 
Univ of Alberta   (he also appeared at Sing Summit)
reinforcement learning book;
AGI is a brash goal;
 
AGI is a group of outsiders...
arrogance and ambition;  eg  isaac newton and gravity  (pointing out the obvious)
 
air is invisible  ... someone had to wake up and say that it was there and had properties
 
Harry Klopf was the grandfather of reinforcement learning  (no one realized it was missing)
 
there is a science of mind... what is MIND?  as in Marr....
 
RL = reinforcement learning...
 
Toward Learning Human-level predictive knowledge
 
every choice is high level  (shall I go to this conference) and low level (wlaking to car)
 
much of the mind is prediction;   modeling the world
world knowledge:   math is timeless and outside the world
world know.  must be predictive....
the world is a source of bits and when we act we create bits;
 
all  scientific know.  must be predictive (of some experiment);
 
 
 
 
 
 </p>