--- Richard Loosemore <[EMAIL PROTECTED]> wrote: > Matt Mahoney wrote: > > --- James Ratcliff <[EMAIL PROTECTED]> wrote: > > > >> Do you not agree that we can achieve AGI without massive compression? > >> Given a large enough storage size? > > > > No, storage has nothing to do with it. My claim is that AGI (on a > > deterministic machine) + some simple code = capability to compress text > and > > video better than any technology that now exists. The existence of such > > compression, proved by a simple test, would prove the existence of AGI. > > Would an AGI with exactly my (human) intelligence be able to pass your > compression test?
Only if your intelligence was uploaded to a deterministic machine. The human brain is not deterministic. To convert a text-based AGI to a compressor, you add an arithmetic coder (about 50 lines of code) and have it predict a text stream. To decompress, you have to reset the AGI back to exactly its previous state (not possible with human brains) and repeat the predictions. To convert an AGI with vision to a video compressor, you extract the visual features (the lossy part), then predict and compress the features (the lossless part) as with text. Decompression requires the extra step of inverting the feature extraction: finding images that generate the same feature vector, that are perceptually the same to the AGI. But I can't prove that this isn't hard. I only claim that if you solve video compression, then you solve vision. -- Matt Mahoney, [EMAIL PROTECTED] ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=fabd7936
