Project

General

Profile

SG separation challenge » History » Version 12

Eli Rykoff, 11/25/2013 12:15 PM

1 1 Ignacio Sevilla
h1. SG separation challenge
2 1 Ignacio Sevilla
3 1 Ignacio Sevilla
Now that several people are testing their own approaches:
4 1 Ignacio Sevilla
5 1 Ignacio Sevilla
* Cut-based with DESDM info (Eli, Diego, Nacho, Ryan...).
6 1 Ignacio Sevilla
* Multi-class (Maayane)
7 1 Ignacio Sevilla
* Random Forests (Ryan)
8 1 Ignacio Sevilla
* Boosted Decision Trees (Nacho)
9 1 Ignacio Sevilla
* Alternative Neural Network with probabilistic output (Chris Bonnett).
10 5 Ignacio Sevilla
* others...
11 1 Ignacio Sevilla
12 1 Ignacio Sevilla
I think the time is right and the codes are mature to launch a specific SG separation challenge, mimicking the successful photo-z WG exercise.
13 1 Ignacio Sevilla
14 1 Ignacio Sevilla
We have to establish:
15 1 Ignacio Sevilla
16 1 Ignacio Sevilla
* The training/validation/testing sample (COSMOS, others).
17 6 Ignacio Sevilla
* Only stars and galaxies? What about QSOs, image artifacts?
18 3 Ignacio Sevilla
* The metrics (Fixed cut, Fixed purity, Fixed Efficiency, ROC -- see example below).
19 7 Eli Rykoff
* SVA1 systematics: correlations with depth, Galactic latitude, seeing, etc.
20 4 Ignacio Sevilla
* Who/how to run it.
21 8 Ignacio Sevilla
* Is there any gain combining them (a committee)?
22 1 Ignacio Sevilla
* The schedule.
23 9 Maayane Soumagnac
24 9 Maayane Soumagnac
h1. The metric
25 9 Maayane Soumagnac
26 9 Maayane Soumagnac
We suggest to use the same metric as in the DES star/galaxy separation (on simulation) paper (arXiv:1306.5236).
27 9 Maayane Soumagnac
28 10 Maayane Soumagnac
h2. Completeness and Purity provided by a given classifier
29 10 Maayane Soumagnac
30 9 Maayane Soumagnac
We define the parameters used to quantify the quality of a star/galaxy classifier. For a given class of objects, X (stars or galaxies), we distinguish the surface density of well classified ob jects, N_X , and the misclassified objects, M_X .
31 9 Maayane Soumagnac
32 9 Maayane Soumagnac
* The galaxy completeness c^g is defined as the ratio of the number of true galaxies classified as galaxies to the total number of true galaxies. 
33 9 Maayane Soumagnac
* The stellar contamination f_s is defined as the ratio of stars classified as galaxies to the total amount of ob jects classified as galaxies. 
34 9 Maayane Soumagnac
* The purity p^g is defined as 1-f_s
35 9 Maayane Soumagnac
36 11 Maayane Soumagnac
!{width:400px}metric.png! 
37 10 Maayane Soumagnac
38 10 Maayane Soumagnac
h2. Plots
39 10 Maayane Soumagnac
40 10 Maayane Soumagnac
Bellow are three different plots we suggest to use to assess the performances of each classifier.
41 10 Maayane Soumagnac
42 10 Maayane Soumagnac
h3. Histograms
43 10 Maayane Soumagnac
44 10 Maayane Soumagnac
Example, on simulations, from arXiv:1306.5236
45 11 Maayane Soumagnac
!{width:400px}histo.png!
46 10 Maayane Soumagnac
47 10 Maayane Soumagnac
h3. purity as a function of magnitude (for fixed completeness, the threshold/cut is let free)
48 1 Ignacio Sevilla
49 12 Eli Rykoff
!{width:800px}Emma.png!
50 10 Maayane Soumagnac
51 10 Maayane Soumagnac
h3. completeness as a function of magnitude (for fixed completeness, the threshold/cut is let free )