Author Topic: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)  (Read 18709 times)

jules

  • Hero Member
  • *****
  • Posts: 8402
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #30 on: May 20, 2012, 01:30:35 pm »
Yes - you're right! There are user-controlled browser and settings related differences which can't be avoided. Ah well - as consistent as possible then!! ;)

Interesting experiment you have developed here! Looking forward to the results. :D 8)

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #31 on: May 20, 2012, 02:27:10 pm »
Thanks jules. Yes, whatever they are, the results will be interesting!  ;D

I'm delighted that fellow zooites are also interested in this project!  8)

I've sent a third lot of candidates out, by PM. If you didn't get a PM from me, and would like a chance to classify them, please send me a PM.


JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #32 on: May 20, 2012, 05:26:40 pm »
For comparison, here is a giant elliptical (same 1 pixel = 900 pc scale, 36 kpc circle added):



SDSS J020328.70-094900.0 (DR7 ObjId 587727178462789849), z=0.322

I've begun my third CasJobs; here are a couple of early finds:



SDSS J233428.47+002800.8 (DR7 ObjId 588015509803434248), z=0.426 (left; posted before by gumbosea, in the possible strong gravitational lenses thread)
SDSS J084236.43+292539.1 (DR7 ObjId 587735042539127126), z=0.416 (right; posted at least four times before, first - as far as I can tell - by gz, in the Ring Galaxy thread)

paulrogers

  • Guest
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #33 on: May 20, 2012, 05:41:18 pm »
Anyway, for what I'm trying to do here*, I'd like the long-term zooites who are having a go at classifying - visually! re morphology! - these "80 pixel fuzzballs" to give it their best shot. In other words, this is my 'committee of experts', partly because many of you are better at this than I am.

In classifying I have always insisted on clear evidence of a spiral structure, not just impressions or hints.  If it's a "fuzzball", that doesn't cut it.  I've seen times when there's just a hint of slightly streaky darkening in an image, but I've rejected it as evidence of an arm like structure, knowing that there could be intervening gas clouds.  For me it'd have to be a pretty sharp 80 pixels!  Fuzzballs need not apply.

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #34 on: May 20, 2012, 06:01:10 pm »
Anyway, for what I'm trying to do here*, I'd like the long-term zooites who are having a go at classifying - visually! re morphology! - these "80 pixel fuzzballs" to give it their best shot. In other words, this is my 'committee of experts', partly because many of you are better at this than I am.

In classifying I have always insisted on clear evidence of a spiral structure, not just impressions or hints.  If it's a "fuzzball", that doesn't cut it.  I've seen times when there's just a hint of slightly streaky darkening in an image, but I've rejected it as evidence of an arm like structure, knowing that there could be intervening gas clouds.  For me it'd have to be a pretty sharp 80 pixels!  Fuzzballs need not apply.

Thanks!

Of the six images I've posted so far, in this thread, of z>0.3 spiral galaxy candidates, which would you classify as spirals (if you don't mind me asking)?

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #35 on: May 20, 2012, 06:32:41 pm »
Oh, and I forgot to add: in this particular mini-project, 'spiral galaxy' has a somewhat non-standard definition.

Yes, galaxies with clearly identifiable arms are spirals. So are, for my mini-project, "flocculent spirals" - "The primary characteristic of flocculent spiral galaxies is the disjoint nature of the individual arms. The prototype is NGC 2841, ..." (source; check out this set of Hubble images of NGC 2841) - barred spirals (even if the arms aren't easy to see), edge-on (or nearly edge-on) spirals, ring and ringed galaxies (but not polar ring galaxies), ... in fact, almost every kind of galaxy except ellipticals!  :o Just about the only other exceptions are obvious mergers, and overlaps (where the morphology of the z>0.3 galaxy - whether foreground or background - can't be ascertained, because of the overlap).

Unfortunately, the scope and objectives of this mini-project are rather, um, fluid  :(

paulrogers

  • Guest
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #36 on: May 20, 2012, 08:39:32 pm »
Of the six images I've posted so far, in this thread, of z>0.3 spiral galaxy candidates, which would you classify as spirals (if you don't mind me asking)?

Only the last, provisionally.  I don't absolutely have to see arms, because these elongated ovals are not stable unless they have angular momentum, i.e. are tilted spirals.  At some distant point, we just can't see well enough to see the arms.  Are there non-spiral ellipticals that are severely flattened like that?  Well, perhaps, but I'd rather bet on tipped spirals.

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #37 on: May 20, 2012, 10:06:52 pm »
Anyway, for what I'm trying to do here*, I'd like the long-term zooites who are having a go at classifying - visually! re morphology! - these "80 pixel fuzzballs" to give it their best shot. In other words, this is my 'committee of experts', partly because many of you are better at this than I am.

In classifying I have always insisted on clear evidence of a spiral structure, not just impressions or hints.  If it's a "fuzzball", that doesn't cut it.  I've seen times when there's just a hint of slightly streaky darkening in an image, but I've rejected it as evidence of an arm like structure, knowing that there could be intervening gas clouds.  For me it'd have to be a pretty sharp 80 pixels!  Fuzzballs need not apply.
Oh dear, a serious mis-communication that I was totally blind to  :( :'(

The fuzzballs are ~80 pixels wide, or ~40 pixels in radius; that means the total number of 'spatial signal pixels' is ~5,000. And as they're RGB pixels, that means ~15,000  8)

I once heard that Fritz Zwicky* could glance at a photographic plate - which is, of course, monochrome - see a tiny blob, and declare it to be a spiral (or elliptical, or ...). If you were to examine the plate with a microscope, you might find that the blob was comprised of ~ten grains of silver**. An 80 pixel wide fuzzball is a veritable cornucopia of data compared with ten grains of silver ...

* or was it Allan Sandage?
** astronomers used glass plates covered with a photographic emulsion, the key ingredient in it being a silver halide (or a combo of silver halides). When hit by a photon (or hundred) the Ag-Cl (say) bond in a crystal would break, releasing the chlorine. In developing the plate, grains in which some AgCl molecules had been zapped by photons would be converted entirely to silver metal; these would be the specks on the plate.

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #38 on: May 21, 2012, 09:39:24 am »
Here's another "run 3" find (I've already posted it in the possible strong gravitational lens thread):



SDSS J095917.01+010159.0 (DR7 ObjId 587727942423216392), z=0.446.


But even here, in GZ classifications, we can choose to invert the images we're presented. And no one has control over each zooite's browser settings (well, other than the zooites themselves!  ;) ). In fact, if I recall correctly, in Ice Hunters (or was it Moon Zoo?) the instructions (or guide, in the forum) explicitly includes advice on how to modify the parameters of the browser image display.

Anyway, for what I'm trying to do here*, I'd like the long-term zooites who are having a go at classifying - visually! re morphology! - these "80 pixel wide fuzzballs" to give it their best shot. In other words, this is my 'committee of experts', partly because many of you are better at this than I am.

* well, today; this mini-project has undergone many twists and turns  :P
(The bolding is not in the original, and I added the bit about the diameter of the fuzzball  ;))

I'm reading "Thinking, Fast and Slow", by Daniel Kahneman, and remembered reading this passage:
Quote from: Kahneman
We have all heard such stories of expert intuition: the chess master who walks past a street game and announces "White mates in three" without stopping, or the physician who makes a complex diagnosis after a single glance at a patient. Expert intuition strikes us as magical, but it is not. [...] The psychology of accurate intuition involves no magic. Perhaps the best short statement of it is by ... who studied chess masters and showed that after thousands of hours of practice they come to see the pieces on the board differently from the rest of us. [...] "The situation has provided a cue; this cue has given the expert access to information stored in memory, and the information provides the answer. Intuition is nothing more and nothing less than recognition."

Here too are a few snippets from jules' excellent blog entry on the first Zooniverse Project Workshop (context is important; I urge you to read the whole thing):
Quote from: jules
Some of the Galaxy Zoo images were too indistinct for computers to sort reliably. ... Classifications need to be good. Quality as well as quantity is needed. Users were weighted by matching with expert classification (Kevin.)

The advantages are: ... Leaves room for other interpretations – ie doesn’t confine people to classifying in a certain way. A rigid tutorial could result in predefined answers.

Issues to consider: User modelling / user performance – some people are better than others at classifying

Basic idea: ... weight users, include some expert measurements, make the casual count – ie a single classification is important (Most people on Galaxy Zoo do a couple of clicks and then leave.) Half the data are from users spending less than 1 minute classifying

From the classifications I've received so far, of the (now) ten galaxies (distant, giant spiral candidates), I would say there are some with Zwicky/Sandage-like expertise!  :o  ;D  8)  8)

Thank you very much, fellow zooites!  :-*

Blackprojects

  • Hero Member
  • *****
  • Posts: 9518
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #39 on: May 22, 2012, 12:55:02 am »
ZOOITESThose with Several 1,000 Classifactions or More are Probably the People who can answer all the questions as we Devellop a Sense of what a GALAXY is Just by a Quick look!

Some of us have gone onto Other Zoo"s and Hardly spend any time on Hubble Zoo any more my self i am a Dedicated Super Novae ZOOITE  With 15,325 Sorted so Far on SN ZOO Compared with 945 on Hubble Zoo and 4,636 on MoonZoo! ZOO2 and ZOO1 Had a vast amount of Clicks on them!

Yet i am Still only a Begginer Compared with Some of my fellow ZOOITES! Clicks wise!

graham d

  • Hero Member
  • *****
  • Posts: 2408
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #40 on: May 22, 2012, 07:50:10 pm »
Quote
ie a single classification is important (Most people on Galaxy Zoo do a couple of clicks and then leave.) Half the data are from users spending less than 1 minute classifying
This says it all. However, if a few clicks or less than a minute classifying means something important please explain.
Quote
Perhaps the best short statement of it is by ... who studied chess masters and showed that after thousands of hours of practice they come to see the pieces on the board differently from the rest of us. [...] "The situation has provided a cue; this cue has given the expert access to information stored in memory, and the information provides the answer. Intuition is nothing more and nothing less than recognition."
A chess master doesn't become a master by chance. As well as innate ability there's constant practice and theoretical study. Infact, for a while the more books one studies the worse ones game gets. This would be philosopher must have been drunk when he wrote that final line.

As for looking at emulsions and chasing the determination of Hubble's constant Sandage appears to have creating two warring factions. Maybe Bill or someone could say some kind words about them both. Zwicky despite his brilliance was pretty un American in his attitude to spherical cows.

Those z>0.3 fuzzballs; you can't beat higher resolution images can you? Hubble zoo images are superb for a while, but we need even more resolution.

 Why choose that lifeless rock; I think I managed a few clicks on it. Compared with the whole heavens surround , why choose that light polluting mass?

As for weather data, some filled in by booze crazed pirates, and ancient Greek classifications. Let them set up their own sites. It's quicker to learn classical Greek and decipher the whole in seconds. What we would all like to know is why Cosmoquest got the boot?
« Last Edit: May 22, 2012, 07:58:41 pm by graham d »

Blackprojects

  • Hero Member
  • *****
  • Posts: 9518
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #41 on: May 22, 2012, 10:19:43 pm »
Graham that may be Some thing for a chat in a Pub to Theorise over but not really for here! :o :o :o :o

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #42 on: May 31, 2012, 09:49:10 am »
Quote from: JeanTate
Time for a history-of-astronomy trivia quiz question: you have all heard, I'm sure, of the fact that some people - even some astronomers - do not accept that the Hubble distance-redshift relationship holds for all galaxies (once you take account of some small, local variations); what role did M81 and NGC 309 play, as evidence against this relationship? For an extra bonus point, what is the name of the astronomer made the connection? For two bonus points, what's wrong with this evidence (that involving M81 and NGC 309)? I'll post answers later.  :D  :-*

Yes, I did say that, back in the very first post, on page one. No answers from me yet, despite my promise.  :( Well, here's a start ...



This is Figure 18, from a book by Halton Arp; the caption reads "At its supposed redshift distance, the ScI galaxy NGC 309 is compared to M81, one of the largest galaxies of whose dimensions we can be sure. NGC 309 would be so enormous that it would make M81 look like a knot in one of its spiral arms." (source)

I don't know when this was written, but here's my version of this same figure, using distances from NED (and images from SDSS  ;D)



While the two galaxies are still quite different in (estimated) actual size, the disparity has shrunk some; no doubt this is due in large part to better, and more accurate, estimates of the distances to these two galaxies. Estimates which do not rely upon their redshifts, "Redshift-independent distances", as NED puts it.

Zooites have posted many hundred (thousand, actually) examples of Galaxy pairs which overlap but are not merging. Many of these are of galaxies with similar redshifts, but which do not appear to be interacting. One such pair is IC 1252 and companion (I don't know what this companion is called, other than SDSS J171547.77+572209.2)*; they have redshifts of 0.0300 and 0.0319. Look how different in size they appear!  :o 8)



One more thing: other than that it uses the logical fallacy "argument from incredulity" (or appears to use that fallacy), what's wrong with the evidence presented in Figure 18?

* strangely, I could find no post, by any zooite, of either, using their DR7 ObjIds, in this forum; perhaps they're posted using different identifiers? Thanks to NGC3314 for bringing this pair to my attention.

NGC3314

  • Global Moderator
  • Hero Member
  • ****
  • Posts: 1590
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #43 on: May 31, 2012, 06:18:25 pm »

* strangely, I could find no post, by any zooite, of either, using their DR7 ObjIds, in this forum; perhaps they're posted using different identifiers? Thanks to NGC3314 for bringing this pair to my attention.

Not that it's much less mysterious, but my detailed document (one of the PDFs  at data.galaxyzoo.org/overlaps.html) says that I picked it up from the forum but outside the overlap thread on 28 April 2009.

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Wednesday, 16th May, 2012: Seeing Is Believing (Or Is It?)
« Reply #44 on: June 03, 2012, 07:45:06 pm »
A VERY BIG THANK YOU for all my fellow zooites who provided me with your classifications of the candidates!  :-*  8)  ;D

In Galaxy Zoo 1 : Data Release of Morphological Classifications for nearly 900,000 galaxies - the paper which presents the results of the millions of zooite clicks, in the original Galaxy Zoo - there are three sets of galaxies*: superclean, clean, and greater. What does this mean? A galaxy in the clean set had "at least 80 per cent of the vote [...] assigned a classification"; in other words, if >80% of the zooite classification clicks for a galaxy were "Elliptical galaxy", this galaxy would be in the clean set. The threshold for a galaxy to be in the superclean set is 95%, and for the greater set, 51%.

The original Galaxy Zoo collected ~30+ votes (clicks) for every object; in my 'giant, distant spirals' project I count success as getting 10+ independent classifications. I've now got at least that number for ~half my candidates, and am close to getting 10+ for the other half (if you'd like to have a go at classifying, and haven't received a PM from my, with an invitation, well, you're more than welcome to join in!  :D Just send me a PM  ;)).

Four of my candidates are superclean; in fact, all zooites who classified these voted 'spiral'.  8) Here they are, together with my tracking numbers:

02: 03:
05: 21:

Here are some of the additional comments on each:

02: Anti-clockwise; disrupted; not 'regular'
03: (Nearly) face-on; with a slight tilt; anti-clockwise
05: Somewhat disheveled; disturbed; fuzzy; at least two fairly tight arms
21: Face-on; anti-clockwise; two-armed; with an oval bar

There are five others which belong to the clean set (but not the superclean one); interestingly, they are all classified as spirals.

The galaxy with the greatest proportion of votes for 'elliptical' belongs to the greater set:

17:

(more later).

Oh, for completeness, I should add that these results are preliminary; I need to check everything once more, to make sure I haven't made any mistakes.

* I'm referring to those in Table 2 of that paper