Slashdot Mirror
Identify Galaxies Using Spare Wetware Cycles
hazem invites us to have fun, learn about galaxies, and actually help astronomers by looking at pictures of galaxies and identifying the type. Warning: it's more addictive than Tetris. From the site: "GalaxyZoo... harnesses the power of the internet — and your brain — to classify a million galaxies. By taking part, you'll not only be contributing to scientific research, but you'll view parts of the Universe that literally no-one has ever seen before and get a sense of the glorious diversity of galaxies that pepper the sky. Why do we need you? The simple answer is that the human brain is much better at recognizing patterns than a computer can ever be. Any computer program we write to sort our galaxies into categories would do a reasonable job, but it would also inevitably throw out the unusual, the weird and the wonderful. To rescue these interesting systems which have a story to tell, we need you."
is this going to work out anything like google image tag game did? so people classify these galaxies and with like 3 or 4 classifying the same galaxy, seeing which tags/classifications are agreed upon?
Sigs are too short to say anything truly profound so read the above post instead.
"Tetris Diary: Day One. This will be an ongoing catalog of the various Tetris shapes I see while playing the game.
First: A cube. Good start!
Second: A clockwise L-shape. I can feel the tension mounting!
Third: A counter-clockwise L-shape. What are the odds??
Fourth: A counter-clockwise S-shape! A trend emerges!
Fifth: A clockwise S-shape. Unbelievable!
Sixth: A STRAIGHT LINE! WE HAVE A STRAIGHT LINE!!!!
I have now reached the top of the screen and the game has ended. Will start again and try to contain my unbelievable excitement over cataloging shapes."
Lawrence Person (lawrencepersonh@gmailh.com (remove all "h"s to mail)
http://www.lawrenceperson.com/
The problem is that computer programs can easily mis-identify things. You should try going over a few of the galaxies they present.. not the tutorial ones.. those are easy.. the actual ones. They're mostly extremely vague, low-res, pixelated jpeg-style-artifacty blobs. In fact, on most of them, I'm having to click "don't know", but on some of them, very vaguely, I can see a spiral.. but to a computer program - that would still be a blob. On the opposite end.. what we clearly identify as a merger, a computer program might think it to be a funky spiral.
That said.. as I mentioned.. most of the actual images are pretty much unidentifiable.. it would be nice if they would concentrate on getting higher resolution images first.. it would make identification easier and more robust.
I understand that maybe they can't.. but a database full of "don't know"-unrecognizable blobs.. I'm not sure what the value is.
Such a project only makes sense if there are a lot of galaxies. And indeed there are: thousands are visible, and estimates of the grand total vary between 100 billion and half a trillion.
Big numbers. But don't forget that each galaxy contains hundreds of millions of stars. Of which ours is just one.
Which should give us all a little humility. But it won't.
NewScientist Article:
http://space.newscientist.com/article/dn12241-pub
Additional Background info here, linked to from that article:
http://space.newscientist.com/article/mg19425994.
Compare this to the Space.com - AP Article:
http://www.space.com/scienceastronomy/070711_ap_o
For whatever reason, the article that Space.com decided to go with fails to mention anything about this project representing a threat to mainstream cosmology or the CMB. Astrophysical enthusiasts reading Space.com, in other words, would not be informed by that article that somebody has even alleged that there is a possible anomalous artifact within the cosmic microwave background. I'm not advocating anything here other than that this appears to be more than a mere "dumbing down" of a complicated story. They could have easily dumbed down the concept of aligned galaxies and why that introduces a problem for the CMB. Instead, we got the following, which appears to not suggest any threat level to BB Theory whatsoever:
This sort of "damage control", if I may call it that, is not really very helpful when it comes to layman trying to understand what to believe.
We must be very careful of how we promote certain sceintific theories over others. It would be very easy to create a false consensus within society using public relations in this way.
"A man cannot begin to learn that which he thinks he already knows." --Epictetus, 1st Century A.D.
funny? more like awesome. computers can do certain stuff super well, but when it comes to a lot of things, they sputter and die. image recognition is going to be one of those things that computers don't do well for many many years.
feels good not to be obsolete. yet.
Beware the Jubjub bird, and shun the frumious Bandersnatch.
FTFA: The simple answer is that the human brain is much better at recognizing patterns than a computer can ever be. Any computer program we write to sort our galaxies into categories would do a reasonable job, but it would also inevitably throw out the unusual, the weird and the wonderful.
You can't target this at geeks and not a get a weird grin. Computers actually could recognize those galaxies fine, AND mark the unusual, weird and wonderful for additional review. It's a matter of putting in a simple threshold of matching features when you analyze the patterns.
computers can do certain stuff super well, but when it comes to a lot of things, they sputter and die. image recognition is going to be one of those things that computers don't do well for many many years.
feels good not to be obsolete. yet.
Feel good while you can, we've been around for millions of years, and computers have been around for around 50 years, and we're already going into multi-core hardware. Sooner than later, massively parallel hardware patterns will emerge, and coding super-fast neural networks in those will be a child's play. All that's left at this point, would be training the computers to do what you want them to do, like you would a little child.
I did this as a summer job. The good news: it put my name on two physics papers. The bad news: it's boring as hell.
They should occasionally display the "Goatse Nebula" just to keep people awake.
Table-ized A.I.