Forum Home | User Profile | Register | Members | Groups | Search | FAQ | Back to:OnIntelligence.org
onintelligence.org Forum Index -> Models and Simulation Topics -> Learning in HTM?

 
This topic is locked: you cannot edit posts or make replies.   This forum is locked: you cannot post, reply to, or edit topics. View previous topic :: View next topic  


Author Message
Jayman


Joined: 27 Feb 2006
Posts: 4

01-24-07, 04:29 am
PostPost subject: Learning in HTM? Reply with quote

Hi,

After reading a little about HTM I'm still not clear on how one can go about directing the learning process for a htm, i.e. How would you go about effectively implementing a reward based learning mechanism?
Back to top
View user's profile Send private message

Author Message
wlorenz65


Joined: 24 Jul 2008
Posts: 4

07-24-08, 11:46 pm
PostPost subject: Reply with quote

Reward based learning is one of the key concepts of strong AI and strong AIs would make HTM programmers superfluous.

Jeff is one of the owners of Numenta which gets money from HTM programmers. So he will do everything he can to prevent the implementation of reward based learning in HTMs, or at least delay it.
Back to top
View user's profile Send private message

Author Message
Lloyd


Joined: 13 Apr 2007
Posts: 11

09-14-08, 08:57 pm
PostPost subject: Reward Reply with quote

W., aren't you being a bit harsh on strong AI? after all, rewards in the general sense are just a way for the pattern recognition system to know whether or not to keep the latest patterns. It seems to me that the concept would apply to an HTM-type system as well as to a more traditional AI approach.
Back to top
View user's profile Send private message Send e-mail

Author Message
wlorenz65


Joined: 24 Jul 2008
Posts: 4

09-15-08, 09:41 pm
PostPost subject: Re: Reward Reply with quote

Lloyd wrote:
rewards in the general sense are just a way for the pattern recognition system to know whether or not to keep the latest patterns.

Throwing away all bad experiences and running into the same mess over and over again? Not very clever.

Quote:
It seems to me that the concept would apply to an HTM-type system as well as to a more traditional AI approach.

The reward concept applies to technics in general. The only difference is that strong AIs can learn on their own, while narrow AIs need a human to be included in the developement process.
_________________
Sick of getting "programming" when you ordered "learning"? See http://home.arcor.de/w.lorenz65/mlbench for a way to unmask the cheating.
Back to top
View user's profile Send private message

Display posts from previous:   
This topic is locked: you cannot edit posts or make replies.   This forum is locked: you cannot post, reply to, or edit topics.    Page 1 of 1 All times are GMT - 8 Hours

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2002 phpBB Group

Please contact the board administrators if you have any questions regarding the OnIntelligence.org forums.