J/ApJ/885/85 Automated DASH classification for supernovae (Muthukrishna+, 2019)
DASH: Deep learning for the Automated spectral classification of Supernovae and
their Hosts.
Muthukrishna D., Parkinson D., Tucker B.E.
<Astrophys. J., 885, 85-85 (2019)>
=2019ApJ...885...85M 2019ApJ...885...85M (SIMBAD/NED BibCode)
ADC_Keywords: Supernovae; Redshifts; Spectroscopy; Optical
Keywords: methods: data analysis; methods: statistical; supernovae: general
surveys; techniques: spectroscopic
Abstract:
We present DASH (Deep Automated Supernova and Host classifier), a
novel software package that automates the classification of the type,
age, redshift, and host galaxy of supernova spectra. DASH makes use of
a new approach that does not rely on iterative template-matching
techniques like all previous software, but instead classifies based on
the learned features of each supernova's type and age. It has achieved
this by employing a deep convolutional neural network to train a
matching algorithm. This approach has enabled DASH to be orders of
magnitude faster than previous tools, being able to accurately
classify hundreds or thousands of objects within seconds. We have
tested its performance on 4yr of data from the Australian Dark Energy
Survey (OzDES). The deep learning models were developed using
TensorFlow and were trained using over 4000 supernova spectra taken
from the CfA Supernova Program and the Berkeley SN Ia Program as used
in SNID (Supernova Identification software). Unlike template-matching
methods, the trained models are independent of the number of spectra
in the training data, which allows for DASH's unprecedented speed. We
have developed both a graphical interface for easy visual
classification and analysis of supernovae and a Python library for the
autonomous and quick classification of several supernova spectra. The
speed, accuracy, user-friendliness, and versatility of DASH present an
advancement to existing spectral classification tools. We have made
the code publicly available on GitHub and PyPI (pip install astrodash)
to allow for further contributions and development. The package
documentation is available at http://astrodash.readthedocs.io/
Description:
We collected labeled spectra from three main repositories: the SNID
database, the Berkeley Supernovae Ia Program (BSNIP), and the releases
from Liu & Modjaz in 2014-2016.
Combining the spectra from the SNID Templates 2.0 database, the Liu &
Modjaz updates, and the BSNIP v7.0 release, and removing spectra with
unknown ages, we were left with a total of 4831 unique spectra across
403 unique SNe.
See Section 2.
In Table 2, we compare DASH classifications to OzDES ATels from 2015
to 2017. This is summarized in section 5.2.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table2.dat 79 212 *Classification of supernovae released in the
past 3yr of ATels by OzDES
--------------------------------------------------------------------------------
Note on table2.dat: Classification of Supernovae Released in the Past 3yr of
ATels by OzDES (Bassett+ 2015ATel.8164....1B 2015ATel.8164....1B ; Davis+ 2015ATel.8367....1D 2015ATel.8367....1D ;
Glazebrook+ 2015ATel.8413....1G 2015ATel.8413....1G ; Lewis+ 2015ATel.8167....1L 2015ATel.8167....1L ;
Pan+ 2015ATel.8460....1P 2015ATel.8460....1P ; Smith+ 2015ATel.8176....1S 2015ATel.8176....1S ;
Tucker+ 2015ATel.8137....1T 2015ATel.8137....1T ; Yuan+ 2015ATel.8464....1Y 2015ATel.8464....1Y ;
Hoormann+ 2016ATel.9855....1H 2016ATel.9855....1H ; King+ 2016ATel.9570....1K 2016ATel.9570....1K ;
Moller+ 2016ATel.8673....1M 2016ATel.8673....1M ; Mudd+ 2016ATel.9742....1M 2016ATel.9742....1M ;
O'Neill+ 2016ATel.9636....1O 2016ATel.9636....1O & 2016ATel.9637....1O 2016ATel.9637....1O ;
Sommer+ 2016ATel.9504....1S 2016ATel.9504....1S ; Muthukrishna+ 2017ATel10759....1M 2017ATel10759....1M ;
Sharp+ 2017ATel.9961....1S 2017ATel.9961....1S ; Calcino+ 2018ATel11146....1C 2018ATel11146....1C and
2018ATel11147....1C 2018ATel11147....1C ; Macaulay+ 2018ATel11148....1M 2018ATel11148....1M)
--------------------------------------------------------------------------------
See also:
J/AJ/135/1598 : Optical spectroscopy of type Ia supernovae (Matheson+, 2008)
J/AJ/143/126 : Spectroscopy of 462 nearby Type Ia supernovae (Blondin+, 2012)
J/MNRAS/425/1789 : Berkeley supernova Ia program. I. (Silverman+, 2012)
J/AJ/147/99 : Spectroscopy of 73 stripped core-collapse SNe (Modjaz+, 2014)
J/ApJ/827/90 : Spectroscopy of SNe Ib, IIb and Ic (Liu+, 2016)
J/ApJ/832/108 : Spectral properties of Type Ic & Ic-bl SNe (Modjaz+, 2016)
J/ApJS/230/20 : Machine learning technique for CoNFIG gal. (Aniyan+, 2017)
J/MNRAS/472/273 : OzDES DR1 (Childress+, 2017)
http://heracles.astro.berkeley.edu/sndb/info : SNDB home page
http://people.lam.fr/blondin.stephane/software/snid/index.html : SNID access
http://github.com/nyusngroup/SESNtemple/tree/master/SNIDtemplates : Liu &
Modjaz templates access
Byte-by-byte Description of file: table2.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 10 A10 --- Name Supernova name (DESYYANaaaa)
12- 16 F5.3 --- z [0.018/0.63] Redshift determined by MARZ
software (Hinton+ 2016A&C....15...61H 2016A&C....15...61H)
18- 38 A21 --- Cl-ATel ATel classification given by OzDES (1)
40- 59 A20 --- Cl-DASH DASH classification
61- 66 F6.4 --- Prob [0.34/1] DASH softmax regression probability
68- 77 A10 --- Rel DASH reliability
79 A1 --- Match? Agreement on the type of the SN by ATel and
DASH? (x=disagree)
--------------------------------------------------------------------------------
Note (1): It details the type and age from maximum. A question mark after the
classification type indicates that the ATel was not certain on the
classification. Most of these ATel classifications were made by the
OzDES team with the help of Superfit or SNID.
--------------------------------------------------------------------------------
History:
From electronic version of the journal
(End) Emmanuelle Perret [CDS] 26-Oct-2023