J/A+A/702/A173 Classification of GRB (Zhu+, 2025)
Unsupervised machine learning classification of gamma-ray bursts based on the
rest-frame prompt emission parameters.
Zhu S.-Y., Shao L., Tam P.-H.T., Zhang F.-W.
<Astron. Astrophys. 702, A173 (2025)>
=2025A&A...702A.173Z 2025A&A...702A.173Z (SIMBAD/NED BibCode)
ADC_Keywords: GRB ; Gamma rays ; Redshifts
Keywords: gamma-ray burst: general
Abstract:
Gamma-ray bursts (GRBs) are generally believed to originate from two
distinct progenitors, compact binary mergers and massive collapsars.
Traditional and some recent machine learning-based classification
schemes predominantly rely on observer-frame physical parameters,
which are significantly affected by the redshift effects and may not
accurately represent the intrinsic properties of GRBs. In particular,
the progenitors usually could only be decided by successful detection
of the multi-band long-term afterglow, which could easily cost days of
devoted effort from multiple global observational utilities. In this
work, we apply the unsupervised machine learning (ML) algorithms
called t-SNE and UMAP to perform GRB classification based on
rest-frame prompt emission parameters. The map results of both t-SNE
and UMAP reveal a clear division of these GRBs into two clusters,
denoted as GRBs-I and GRBs-II. We find that all supernova-associated
GRBs, including the atypical short-duration burst GRB 200826A (now
recognized as collapsar-origin), consistently fall within the GRBs-II
category. Conversely, all kilonova-associated GRBs (except for two
controversial events) are classified as GRBs-I, including the peculiar
long-duration burst GRB 060614 originating from a merger event. In
another words, this clear ML separation of two types of GRBs based
only on prompt properties could correctly predict the results of
progenitors without follow-up afterglow properties. Comparative
analysis with conventional classification methods using T90 and
Ep,z-Eiso correlation demonstrates that our machine learning
approach provides superior discriminative power, particularly in
resolving ambiguous cases of hybrid GRBs.
Description:
We list the prompt emission parameters of the 370 GRBs in the rest
frame. The classification results of T90, EH, EHD, t-SNE and UMAP are
included for easy comparison.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
tablea2.dat 76 370 Rest-frame gamma-ray bursts (GRBs)
--------------------------------------------------------------------------------
Byte-by-byte Description of file: tablea2.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 10 A10 --- GRB Gamma-ray burst name
12- 18 F7.2 s T90 Duration in the observed frame
20 A1 --- S/L [LS] Classification based on T90 (1)
22- 28 F7.5 --- z Redshift
30- 36 F7.2 s T90z Duration in the rest frame
38- 44 F7.2 keV Epz Peak energy in the rest frame
46- 59 F14.8 10-7J Eiso Isotropic energy
61- 62 A2 --- EH [I II] Classification based on the
EH parameter (2)
64- 66 A3 --- EHD [I II] Classification based on the
EHD parameter (2)
68- 69 A2 --- t-SNE [I II] Classification based on the
t-SNE method (3)
71- 72 A2 --- UMAP [I II] Classification based on the
UMAP method (3)
74- 76 A3 --- Assoc [SN MGF KN] Association with other counterpart
--------------------------------------------------------------------------------
Note (1): Classification based on T90 as follows:
L = Long GRB
S = Short GRB
Note (2): Classification based on EH or EHD parameter as follows:
I = Type I GRBs
II = Type II GRBs
Note (3): Classification based on t-SNE or UMAP method as follows:
I = GRBs-I
II = GRBs-II
--------------------------------------------------------------------------------
Acknowledgements:
Si-Yuan Zhu, zhusy37(at)mail2.sysu.edu.cn
(End) Patricia Vannier [CDS] 30-Sep-2025