21 replies to this topic

[Teddybot]

Second video:
http://www.dartmouth...hg/videos2.html



Continued here:
http://www.gamedev.n...sed-algorithms/

Edited by Teddybot, 02 May 2012 - 11:19 AM.

[Teddybot]

willh, on 27 January 2012 - 10:28 AM, said:

wodinoneeye, on 26 January 2012 - 02:24 AM, said:

... the problem becomes MUCH more difficult when you have a dynamic system with temporal effects (when you have to figure out which factors from the past in a sequence of actions were the cause of the results).

This is one of the things that hidden markov models aim to solve. The Viterbi algorithm allows you to determine the most likely sequence of events that could have resulted in a particular outcome.

ANNs may have a bad reputation, but there have been some very interesting developments relating to neural networks recently (see Geoffrey Hinton and U. of Toronto). It's plausible that HTMs are something new and fascinating. If only someone who has used them could explain what they do...... <*cough*teddybot*cough*>

Thank you for the positive viewpoint!
Slower changing events higher in the hierarchy separate small things in time such as motor movements getting up off the couch from the key way points on the way to the fridge. (Temporal Credit Assignment Problem) is specifically addressed in posted pdf. Which invariant representations to reward or punish during and after the task?
Modules around the system make it all happen. Zero modules were published with the HTM psuedocode. Some of the modules operation are mentioned in On Intelligence. It is up to us to innovate with it and develop how it is used. A new paradigm is a great challenge to take on.

Thread continued here:
http://www.gamedev.n...sed-algorithms/

Edited by Teddybot, 02 May 2012 - 11:40 AM.