I think a better tutorial with more examples would greatly help the effort :)
-
by
Vvolodymyr
Hi everyone. I'm not new to Zooniverse, but just tried HiggsHunters project.
These are my fresh off-the-start observations/impressions:
At first it is clear that we are to look for bundles of white* lines that originate off-center, which seems simple enough. But after just a dozen classifications some questions/confusions arise.
-
What if a (w) single white line that certainly doesn't originate at the center is crossing a (c) colored line (that did originate in the center ? Do we mark that vertex?
1a) What if a (w) single white line originating off-center is crossing a (b) blue line that originates from a different off-center point ? do we mark that vertex? -
What if more than 2 (w) white lines cross off center but at a large angle (90 less angle less180) - for example 120 degrees? do we mark that vertex?
-
In the tutorial, there was a slide portraying how a computer "missed that one line further from the center" as opposed to a human observer, but in FAQ it says that we should not mark single lines. This confuses me 😦
-
what if more than 2 (w) lines start off center (and are not touching) - but it looks like they actually could have originated at the center?
-
In graphs with many white lines & more than 2 bundles of white lines, some lines from different bundles will also cross paths - making an impression that that is also a vertex. Do we mark all possible vertices (even though they are mutually exclusive) - or only our best guess vertices (avoiding the mutually exclusive possibilities)?
I understand that a learning curve is intended - but I feel that if I keep on classifying without clarification, I will not learn anything new, or I won't adapt or get better at it, and will simply keep on doing it wrong.
Posted
-
-
by
mike-ds
I'm sure a moderator will get back to you with answers to your questions soon. I am interested in their response too.
Posted
-
by
Vvolodymyr
Yep. I just don't feel like I understand well enough, and would hate to junk up their data because of that lack of understanding.
But I would definitely like to participate ! 😃P.S. As seen in many projects on Zooniverse, it is clear that the researchers don't want to provide stringent directions, so that they could capitalize on the large amount of varied classifications - since they already have automated data processing tools. They need the human variety.
BUT they also need a good* data collection that would arise from that variety (to help train their automated tools, presumably etc.). So we do need clear instructions on, at least, what we definitely shouldn't do. Basically instructions that would prevent us from making completely irrelevant classifications/observations. The simplest thing would be DO's and DON'Ts comparisons for various situations - everything in between is up to our judgement.
Anyways, I'll wait till I reach better confidence in my ability to contribute here 😃
Posted
-
by
Whoandwhatitis
moderator
in response to Vvolodymyr's comment.
Hello @Vvolodymyr, here is my opinion for your questions (I am not a professional in the particle physics field):
- Maybe yes, you can mark these if it looks like an off-center vertex. Typically, you might see a white line cross a colored line within the boundaries of the inner detector, and this could indicate a collision which generates charged particles - and therefore multiple lines (of potentially any color). Keep in mind, we are looking at a 2-dimensional representation of a 3-dimensional detector with respect to time, so this is understandably tricky to decipher.
1a) Maybe yes, once again. The particle represented by the blue line might decay at the point where the blue line and white line meet; or they could be completely unrelated, and the path of one just happened to intersect the other at a different time.
Probably the best way to know is that one you're done with your classification, check out the discussion page and flip through the three available views. Often the Slice view can help you to understand more about the collision than you can find if, for example, you were classifying the Zoom or Normal view. - Probably not, for most objects.
- Do you mean the middle image in the "Multiple" section? That's correct, though - single lines do not represent an off-center vertex.
The computer has spotted the vertex with 5 lines and coloured it in,
but it looks like it’s completely missed the one with 2 lines! This
goes to show how much better the human eye is at this task.- Usually these might not be related to an off-center vertex; they may be part of what we label as a "bundle".
- Please provide some examples, and we can check them out as a community. It sounds like it would be best to mark a little of both, depending on each candidate. A bundle might appear due to the decay of a particle far away from the center, and the computers generated an image which shows a bundle intersecting at this point because they are attempting to draw a line from a detected particle, back to the central point with respect to the collision point.
All of that said, check out this fairly recent blog post which shows the first analysis of our ability as a community to point out the off-center vertices by Will Kalderon:
https://blog.higgshunters.org/2016/10/15/first-results-from-your-clicks-submitted-to-a-journal/Edit: Oh, and typically I mark most bundles as "Something weird".
Posted
- Maybe yes, you can mark these if it looks like an off-center vertex. Typically, you might see a white line cross a colored line within the boundaries of the inner detector, and this could indicate a collision which generates charged particles - and therefore multiple lines (of potentially any color). Keep in mind, we are looking at a 2-dimensional representation of a 3-dimensional detector with respect to time, so this is understandably tricky to decipher.