Numenta's BTW isn't so similar to my work, they look like they only are presenting a framework / idea to be built on, which doesn't really make much sense either. I didn't see anything working or useful they present, was veryyyy disappointed.
Hmm symbol neural logic you say. In my AGI blueprint there is no symbology or logic gates....just merging nodes to make a small accurate brain combustion. The brain I explain can Do logic like AND OR NOT etc, or simply predict a next word or do algebra. And there's no limit to context, you can look at the last 5, and 6, and 9 words to make a prediction, there is no logic based prediction A>B. It's abcdef>g, you have lots of context in the prompt fed in. As for symbols, again this is an illegal word here :). All data is data, text is only cuz we can't share easily our vision mentally, text is not actually needed, but it is for now made a "other" sense to our brains like smell or taste or sight. There is no GOFAI for text based reasoning, it IS neural AGI! I have it all laid out, you can even use image recognition for distortions in text. Currently blackboxes are used for image training, but again this can be the good ol fasion merging into a hierarcy, it is the same mechanisms that make a hierarchy etc for text, same same same, mine uses No backprop. The field has a serious problem with understanding ai using backprop and dispensing text neural network architecture! It is just a simple net pyramid that merges clusters, for both text and image etc. Mine can explain all it's decisions, fully unsupervised net, it'll learn patterns deeper in many ways, no hardcoding much. All that there is is patterns in data, it all starts with Exact matches and stems from there on to indirect relations, categories etc. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Tcf8fbceae6bd0acd-M0a544a7aab9a84acc2fa6227 Delivery options: https://agi.topicbox.com/groups/agi/subscription
