Classification of astrophysical events from gravitational wave signature. (arXiv:2008.06550v2 [astro-ph.HE] UPDATED)
<a href="http://arxiv.org/find/astro-ph/1/au:+Singh_S/0/1/0/all/0/1">Shashwat Singh</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Singh_A/0/1/0/all/0/1">Amitesh Singh</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Prajapati_A/0/1/0/all/0/1">Ankul Prajapati</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Pathak_K/0/1/0/all/0/1">Kamlesh N Pathak</a>
In recent years, improvements in Deep Learning techniques towards
Gravitational Wave astronomy have led to a significant rise in various
classification algorithm development. A few of these algorithms have been
successfully employed to search gravitational waves from binary blackhole
merger events. However, these algorithms still lack success with significant
time duration and further prediction of the merger events parameters. In this
work, we intended to advance the boundaries of deep learning techniques, using
the convolutional neural networks, to go beyond binary classification and
predicting complicated features that possess physical significance. This method
is not a replacement for the already established and thoroughly examined
methods like matched filtering for the detection of gravitational waves but is
an alternative method wherein human interference is minimal. The deep learning
model we present has been trained on 12s of data segment, aimed to predict the
27 features of any LIGO time series data. During training, the maximum accuracy
attained was 90.93%, with a validation accuracy of 89.97%.
In recent years, improvements in Deep Learning techniques towards
Gravitational Wave astronomy have led to a significant rise in various
classification algorithm development. A few of these algorithms have been
successfully employed to search gravitational waves from binary blackhole
merger events. However, these algorithms still lack success with significant
time duration and further prediction of the merger events parameters. In this
work, we intended to advance the boundaries of deep learning techniques, using
the convolutional neural networks, to go beyond binary classification and
predicting complicated features that possess physical significance. This method
is not a replacement for the already established and thoroughly examined
methods like matched filtering for the detection of gravitational waves but is
an alternative method wherein human interference is minimal. The deep learning
model we present has been trained on 12s of data segment, aimed to predict the
27 features of any LIGO time series data. During training, the maximum accuracy
attained was 90.93%, with a validation accuracy of 89.97%.
http://arxiv.org/icons/sfx.gif
