Lol bro... I'll show you how text is same as vision. And yes my code I made from scratch works with just pattern matching, it uses 0-16 letter matches - combing their predictions of next letter, exponential function for which context length is more learnt and for a set of predictions themselves, recency priming more if closer to end of present time, and scoring good but not shown on my project page is hole matching and delay matching and delay prediction ex. it knows wznd or wiind and both match wind, so it knows a y comes after it if see that context. It can predict a letter in advance, because it happens sometimes it comes earlier - never saw before to it to happen. I have left to do some weighting optimization tricks and related context matching, and mirroring ability, and delay or hole pattern.
Vision and Text: Vision and audio and video focuses on lines and when someone starts or end talking or moving (or flashing), which draw attention to these lines (changes), based on relativity. With these lines, we may have a car say. So you see a car outline but half is missing, you can predict it, just like text, if you saw wheels and see a spoke hole cavity, maybe a wheel should go there! If you saw a bird in the last hole, maybe a bird should go there! (maybe it is a car with birds as wheels, !). What really messes up your mind is that vision and audio, even TEXT! all can do pattern of delay, for example we saw a car outline before, but next time we see a BIG car outline, how do we predict lines now? Simple, relative pattern of delay, so if the recognized features are all 10 pixels wider stretched, then the prediction should be made like that too. RL for prediction of data: Unfortunately in data prediction, you need to force it to more likely predict food, cash, cars, AGI, and data, so that it does not predict and ramble on about clothes, golfing, clouds, and so that it hates predicting death, carrots, pain, and tanning beds. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T5b614d3e3bb8e0da-M0767c4c7187ec41d4178844f Delivery options: https://agi.topicbox.com/groups/agi/subscription
