"Sure there are more appropriate lists to talk about compression on,"
"disassociate ourselves from their esteemed company BY KICKING THEM OFF THE DAMN LIST!!!!!!!! ERROR: SARCASM" How bout instead of writing cryptic stories you just say what you think? Me and Matt and JB are compression dudes on this list, and we are of high intellect indeed... So you don't understand why text compression is so spot on? I can explain, just ask. I'm the explainer around here anyway... Since I'm probabilistically sure you are confused, here: ----------------------------------------------------------------------------- Compression is better the higher probability the predictor correctly thinks. To compress well means it understands the data, cat=dog to group words etc. The best compressors are neural models that mix predictions. It can make analogies to losslessly reinvent all the data back from seemingly nothing, and all AI o the internet do this "repairing rejuvenation", and so do cells etc, fetuses predict/make future based on past context. My 120 lines of code algorithm that made 200,000 bytes into losslessly 59,230 bytes is a text generator, a letter predictor using probabilities, learnt Online as it talks to itself storing what it says to itself. It mixes 13 context models and then does Arithmetic Coding. It uses the past text to predict/generate the future until it reaches the end of the file. My algorithm builds a tree from scratch and during mixing will use global model weight, weight threshold squashing, and adaptive threshold based on how many channels are open to it. If I stop its losslessness it generates mostly silly ramblings that were not in the dataset but somewhat make sense. But I will get it as good as the great GPT-2 yet. This works because my algorithm handles future uncertainty with missing data it didn't see yet aka no very likely answer, so it mixes models to get more virtual data to handle uncertainty. My tree uses frequency for phrases to single letter models. And I haven't even started, there's energy, translation, etc etc to do yet that others have paved the way in already and I'm building on it soon. I'm on the way to building REAL AGI here, I'm telling you this is the way to get started. The models are in the tree, it stores text and frequency and can store semantics in the future, the tree is the neural network basically and that's where the pruning and storing of good enough nodes will result in a robust distributed net that is small, fast, and can recognize long unseen strings. I coded it all from scratch, no reusing others code. As for the copy while mutate thing in evolution, AGI must answer questions never seen before, i.e. it uses surrounding context atoms to predict/make the future ground/sentence (babies in the womb, text generators...). During disabling lossless compression / Arithmitic Coding, it can generate unseen futures that are likely based on the context past. Sounds like physics. And that mutates the future correctly while clones the sentence topic being on topic and generates content. Imagine my text predictor/generator wants to predict after "and in the 1600s the captain said the fortress was _" with better probability to compress better, well it has to model the writer's intellect! It can see 100,000 letters back saying after the 1600s the fortress was rebuilt but before then it was being built. so it can see which to use based on the date. It's more of a match thing. To predict new true answers not in the dataset, you are given a question and simply predict the most likely answer based on large diverse data seen previously, which gives you all the truth your eyes could ever see to help you. So in the dataset you want to predict the cure to cancer wrongly, based on their current knowledge, but in the end you can do it better by using more data globally in the dataset from different domains. The AI has to always want certain futures to be seen, this is what steers the prediction in favor of what we want, and that may help compression since all text leads back to the need for survival aka food sex aka shelter games meetings hockey shops TVs books walks data_collection etc. The past, makes the future, using surrounding context in the embryo fetus/sentence said last, to make on topic specialized forecasts predictions generations!! All of evolution is about info duplication and mutation while cloning the topic, like GPT-2. DNA/cells, ideas, humans, cities, AIs. Patterns are found/created and the brain is like a friend network and also like a magnetic, propagating vibrations in fully feedforward aligned networks built from combinational domains to distribute all energy for most power, that's why the most powerful system are larger and fully aligned within. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T7e5f921f8ed8e057-Ma5558a09662d7987451ee780 Delivery options: https://agi.topicbox.com/groups/agi/subscription
