Talk:predictive coding

Various definitions taken from WP

 * "Predictive coding is a [neurobiologically plausible] scheme for inferring the causes of [sensory] input based on minimizing prediction error."
 * "[Linear ]predictive coding (LPC) is a tool [used mostly in audio signal processing and speech processing] for representing the spectral envelope of a digital signal [of speech] in compressed form, using the information of a linear predictive model"
 * 5.1.3 Lossy compression algorithms
 * Linear predictive coding (LPC): lossy compression by representing the spectral envelope of a digital signal of speech in compressed form

Usage from WP

 * our brains use what he calls Predictive coding. It starts with very broad constraints and expectations for the state of the world, and as expectations are met, it makes more detailed predictions (errors lead to new predictions, or learning processes).

linear predictive coding instead?

 * 70 of the 95 articles in which predictive coding occurs in Wikipedia are also article in which linear predictive coding occurs. I have a feeling that predictive coding is an SoP hypernym for the specific three-word coding techniques that are predictive that may not be SoP. DCDuring TALK 23:03, 21 April 2014 (UTC)

RFC discussion: August 2009–April 2014
Clean up, unless it doesn't meet CFI anyway. Mglovesfun (talk) 18:28, 16 August 2009 (UTC)
 * I've moved this out of Requests for cleanup/archive/2009/Unresolved requests in the hope we can resolve it. - -sche (discuss) 03:28, 5 April 2014 (UTC)
 * Clean enough? --kc_kennylau (talk) 09:12, 21 April 2014 (UTC)
 * I don't think this is really good enough yet. I pulled some usage of predicitive coding from WP. I found 4 uses that seemed to contain definitions that I can't quite reconcile to our latest one. I also noted that most uses of predictive coding are actually uses of linear predictive coding. Predictive coding alone seems to be used in a kind of hand-waving, allusive way that seems SoP to me. I'd certainly like to see citations that support our definition. Also our current definition looks like a material alteration of the attempted definition that preceded it, ie, I would have thought that the first had been deleted and another substituted, rather than that the first was "cleaned up". DCDuring TALK 23:11, 21 April 2014 (UTC)
 * I did the "material alteration", because the original def seemed obviously too narrow, or from one domain not covering the whole concept. However, mine was done in a hurry and if anyone has something better than "coding that is predictive" then their input is welcome! Equinox ◑ 00:51, 22 April 2014 (UTC)
 * McGrawHill Sci & Tech Dictionary, the only OneLook reference that has a definition for the term says: "In data compression, a method of coding information in which a sample value is presented as the error term formed by the difference between the sample and its prediction." DCDuring TALK 01:39, 22 April 2014 (UTC)