In his book "How to build a mind", Kurzweil talks about how a so-called pattern recognizer is the central component of his pattern recognition theory of the mind. I would like to test his theory, because of my background in (try not to laugh) Hubbard’s dianetics and his theory of mind. Specifically, I feel that such recognizer setups could be trained, conditioned, and a form of dianetic engrams could be formed. A reactive mind is an easy enough concept to implement.
All jokes aside, it’s a good jumping off point. What I need help with is implementing the recognizer. I understand hidden Markov models are used in this system, but how exactly? These recognizers serially input activation patterns at a lower level, then the degree of activation is sent to a higher level. My question is how they would be "imprinted" with a pattern in the first place, and how an internal HMM would be set up. Furthermore when would the system determine the right time to retrain sections or levels of itself? Finally how would size, weight, and variability be implemented for the recognizer? If you’d rather discuss this privately, my email is mindpixel at proton mail dot com. Thanks! ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Tb77ed1e55dcede23-M268d080456a7d5314b913826 Delivery options: https://agi.topicbox.com/groups
