Talk:Programming by demonstration

Separate article from "Programming by Example"

The distinction between PBD and PBE for robotics and software researchers is not right, PBD is used quite often by researchers in human-computer interaction, for example. (Unsigned comment)

I know of Programming by example only via software. I was at U.C. Berkeley when PhD candidate Dan Halbert, a student of Susan Graham's, was doing his thesis work on PBE at Xerox PARC. (I think he also did his masters work on that topic.) As I noted in a recent edit here, PbE goes back at least to the mid-80s as a term in software research. PbE as a concept -- well, read the Halbert dissertation I just linked. IIRC, he traces it back quite a few years earlier, at least as far back as some UI work by Warren Teitelman.

If anything, I'd prefer separate PbE/PbD articles, with a note in each that there seems to be some terminology overlap between the robotics uses of PbE/PbD (which as an idea must certainly go back at least as far as Kurt Vonnegut's Player Piano?) and the user-interface support idea of PbE Classic (Halbert/PARC). If you're coming from robotics, look at it this way: maybe PbD is very useful in robotics, so why would you want it associated in any way whatsoever with Microsoft Bob, or Clippy, or that incredibly irritating dolphin that used to swim out at you in Windows (all of which were attempts to make PbE work "for the masses", in Halbert's original vision)? On the other hand, if the two terms really are used interchangeably in both UI work and in robotics, I guess there should be one article after all. :-( Yakushima (talk) 15:21, 26 May 2008 (UTC)

If the approaches in PbE are similar to those in PbD, then there should be only one article about them. The concepts themselves only differ in their application areas. The age of the terms and who uses them should not decide whether there are two articles or one. -Pgan002 (talk) 05:03, 7 December 2008 (UTC)

Couldn't agree more - and they don't. Merge=wrong. Notice removed. Maury Markowitz (talk) 13:21, 6 January 2009 (UTC)

Early work on programing by demonstration

The article states that it dates to mid 80s, but there is work even on the late 70s. See, for instance, the work of Michael Bauer (Programming by Examples) in the AI Journal, published in 1979. I was wondering if, since this article focuses mainly in robotics, if the more theoretical work by Bauer should be considered. —Preceding unsigned comment added by 83.53.136.210 (talk) 10:46, 19 June 2010 (UTC)

Robotics attention needed

• Copyedit
• Refs
• Check facts and ensure no improper statements

Chaosdruid (talk) 20:18, 18 February 2011 (UTC)

Recent work on demonstration collection.

In the space of robot learning, there's been many new platforms made to control robot arms, humanoid / bimanual robots etc. in order to collect demonstrations for imitation learning. I will currently update this article, let me know if it would be better suited elsewhere. "Programming" by demonstration doesn't quite encapsulate the purpose of robot learning, but I say it's close enough... Amacalus (talk) 05:57, 21 July 2024 (UTC)