On Friday, November 12, 2021, at 11:39 AM, James Bowery wrote: > Now, having said all that: Yes, the measurement level of abstraction does get > into the economics of computation resources and, yes, it would be nice to > find approaches that obviate all of the above "incompletenesses", but you > must do better than to redefine the words "lossy" and "lossless" compression > as that merely hobbles an existing approach to these incompletenesses while > at the same time threatening to hobble their practical applications by > confusing the meanings of words.
But that's the issue that started all this. There are at least two pursuits happening with compression in regards to AGI. One is the pursuit of maximal lossless, shorter description lengths, AGI closely driven by AIXI. I understand that since I was involved in the 1990’s of pursuing maximal when attempting to preempt Stacker Electronics compression product. If I could carve out more time today, I would pursue maximal. What I’m running into now all over the place is hybrid compression in AGI theory and engineering. Thus, creating new labels for the thing, lossylosslessness, losslesslossyness, etc. Existing classes of compressors like “Visually Lossless”, “Near Lossless”, “Almost Lossless”, “Compressive Sensing”, etc. are not doing it justice. In AGI I see hybrid everywhere, in agent-to-agent communications, agent sensory, agent memory, and agent intelligence. Engineers don’t wait for scientific nomenclature especially software engineers we come up with labels. If we get stuck in this regulatory mode of shoving square pegs into round holes for the sake of official terminology nothing gets built. And it’s easy to blanket describe it as all lossy when in reality it’s not. BTW I just thought of another it’s called “Black Box Lossless, Easter Egg Lossy” :) ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T5ff6237e11d945fb-Me316459a3498b38f63cb1e22 Delivery options: https://agi.topicbox.com/groups/agi/subscription
