Galaxy Zoo Talk

New phase of GZ? Recycling SDSS images first classified a year or so ago?

  • JeanTate by JeanTate

    I seem to be getting more than a few SDSS objects like this ( AGZ0003lyo):

    enter image description here

    From the comments, I see that another zooite (Terrance L. Johnson) commented on it nearly a year ago now (June 4 2013), and that it's in vrooje's Possibly Bulgeless collection. From a quick check I just did, among the first few pages of recents, I found lots of cases like this.

    Has Galaxy Zoo entered a new phase? One where some objects classified before - up to two years' ago - are being offered up again?

    Perhaps it's related to the fact that there may have been far more repeats, or duplicates (a zooite gets to classify the same object more than once), than was believed to happen (so the target of 20 independent classifications was not attained), and/or the fact that for a significant fraction of objects the classifications are discordant (i.e. lots of zooites classifying down one arm of the decision tree, and about the same number down another arm). Perhaps it's a dedicated experiment to test machine learning, using software and approaches developed at and after the Workshop on Citizen Science in Astronomy. Perhaps ...

    Does anyone know? Has a Science Team member written about this (and I missed it)? Is another 'fake AGN' study under way?

    Posted

  • Capella05 by Capella05 moderator in response to JeanTate's comment.

    Hi Jean 😃

    I have not heard anything, but then again, I have not been on-line for the last fortnight (laptop died - waiting for replacement).

    Will keep my ears open.

    Posted

  • JeanTate by JeanTate in response to Capella05's comment.

    Thanks! 😃

    I was wondering why you'd suddenly disappeared ... hope you didn't lose too much when your laptop died ...

    Posted

  • Capella05 by Capella05 moderator

    Lost everything.

    Do have some things backed up, but need a pc to retrieve them. I think it was the universe telling me to take a break.

    Posted

  • shocko61 by shocko61

    Hi Capella05 ,commiserations ,been there done that , but I now have an external drive that gets backed up and have recently been saved when my computer spat the dummy , Ian

    Posted

  • Capella05 by Capella05 moderator in response to shocko61's comment.

    Hi Ian 😄

    I did have an external drive to back up - so most of the science stuff is safe (fingers crossed).

    Being far away from home, it was the photos that were more important - and I never thought to back them up 😦

    I will know better in future- but it is good to be back after 3 weeks 😃

    Posted

  • KWillett by KWillett scientist, admin, translator

    Hi everyone,

    The SDSS images have a renewed priority in the queue since we recently finished classifying the UKIDSS images in the sample (longer blog post on this will appear tomorrow). No experiments of any kind are running - we're still gathering the same sets of independent classifications we've used on all the projects to date!

    Both the developers and I have looked carefully at the repeat/duplication question, and haven't found any evidence of that happening to particular users. If there's a particular galaxy you're sure you've seen twice, please let us know and we can check, but we haven't seen it happening in GZ so far.

    • kw

    Posted

  • JeanTate by JeanTate in response to KWillett's comment.

    Thanks Kyle.

    Am looking forward to the longer blog post! 😃

    Posted

  • KWillett by KWillett scientist, admin, translator

    Should be up by now!

    Posted

  • JeanTate by JeanTate in response to KWillett's comment.

    Following up on this:

    Both the developers and I have looked carefully at the repeat/duplication question, and haven't found any evidence of that happening to particular users. If there's a particular galaxy you're sure you've seen twice, please let us know and we can check, but we haven't seen it happening in GZ so far.

    I thought there might be something in the blog post on this, but there wasn't.

    It's not clear, from what you wrote, whether you are referring to repeats/dups currently happening, or historically, since GZ's v2 Talk went live (or even before).

    That repeats/dups happened in GZ1 and GZ2 is certain; both Lintott+ 2008 and Willett+ 2013 [1] say so explicitly.

    Repeats/dups occurred in the Star Date (M83), Radio Galaxy Zoo, and Quench, as well as SpaceWarps. There were aspects of the code in M83 and SpaceWarps which were acknowledged as allowing repeats to occur, but not RGZ or Quench. As recent GZ runs have used the same architecture - and code - as both RGZ and Quench, it would quite remarkable if there had been no recent GZ repeats, wouldn't it?

    The RGZ Talk thread - link above - reports some aspects of the challenges the developers faced, in trying to address the repeats; it also reports that efforts to solve the problem are on-going (as of May 12 2014, the problem had not been solved, though it may be by now).

    Do you and the developers know why repeats were so common in Quench, occurred in RGZ, but are effectively absent in recent versions of GZ?

    [1] there is a discussion of the duplicates in GZ2 in this GZ forum thread: Questions about the GZ2DR paper (Willett et al. 2013)

    Posted