arXiv subjects

Daniel Muthukrishna

Publications and source records attributed to Daniel Muthukrishna.

25 records · Page 2Linked to original sources

arXiv2020

Unsupervised machine learning for transient discovery in Deeper, Wider, Faster light curves

Identification of anomalous light curves within time-domain surveys is often challenging. In addition, with the growing number of wide-field surveys and the volume of data produced exceeding astronomers ability for manual evaluation, outlier and anomaly detection is becoming vital for transient science. We present an unsupervised method for transient discovery using a clustering technique and the Astronomaly package. As proof of concept, we evaluate 85553 minute-cadenced light curves collected over two 1.5 hour periods as part of the Deeper, Wider, Faster program, using two different telescope dithering strategies. By combining the clustering technique HDBSCAN with the isolation forest anomaly detection algorithm via the visual interface of Astronomaly, we are able to rapidly isolate anomalous sources for further analysis. We successfully recover the known variable sources, across a range of catalogues from within the fields, and find a further 7 uncatalogued variables and two stellar flare events, including a rarely observed ultra fast flare (5 minute) from a likely M-dwarf.

astro-ph.IM↗

arXiv2020

Using machine learning for transient classification in searches for gravitational-wave counterparts

The large sky localization regions offered by the gravitational-wave interferometers require efficient follow-up of the many counterpart candidates identified by the wide field-of-view telescopes. Given the restricted telescope time, the creation of prioritized lists of the many identified candidates becomes mandatory. Towards this end, we use \text{\astrorapid}, a multi-band photometric lightcurve classifier, to differentiate between kilonovae, supernovae, and other possible transients. We demonstrate our method on the photometric observations of real events. In addtion, the classification performance is tested on simulated lightcurves, both ideally and realistically sampled. We show that after only a few days of observations of an astronomical object, it is possible to rule out candidates as supernovae and other known transients.

astro-ph.HE↗

arXiv2019

RAPID: Early Classification of Explosive Transients using Deep Learning

We present RAPID (Real-time Automated Photometric IDentification), a novel time-series classification tool capable of automatically identifying transients from within a day of the initial alert, to the full lifetime of a light curve. Using a deep recurrent neural network with Gated Recurrent Units (GRUs), we present the first method specifically designed to provide early classifications of astronomical time-series data, typing 12 different transient classes. Our classifier can process light curves with any phase coverage, and it does not rely on deriving computationally expensive features from the data, making RAPID well-suited for processing the millions of alerts that ongoing and upcoming wide-field surveys such as the Zwicky Transient Facility (ZTF), and the Large Synoptic Survey Telescope (LSST) will produce. The classification accuracy improves over the lifetime of the transient as more photometric data becomes available, and across the 12 transient classes, we obtain an average area under the receiver operating characteristic curve of 0.95 and 0.98 at early and late epochs, respectively. We demonstrate RAPID's ability to effectively provide early classifications of observed transients from the ZTF data stream. We have made RAPID available as an open-source software package (https://astrorapid.readthedocs.io) for machine learning-based alert-brokers to use for the autonomous and quick classification of several thousand light curves within a few seconds.

astro-ph.IM↗

arXiv2019

DASH: Deep Learning for the Automated Spectral Classification of Supernovae and their Hosts

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 four years 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, Blondin & Tonry 2007). 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 presents 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 https://astrodash.readthedocs.io.

astro-ph.IM↗

arXiv2018

The Photometric LSST Astronomical Time-series Classification Challenge (PLAsTiCC): Data set

The Photometric LSST Astronomical Time Series Classification Challenge (PLAsTiCC) is an open data challenge to classify simulated astronomical time-series data in preparation for observations from the Large Synoptic Survey Telescope (LSST), which will achieve first light in 2019 and commence its 10-year main survey in 2022. LSST will revolutionize our understanding of the changing sky, discovering and measuring millions of time-varying objects. In this challenge, we pose the question: how well can we classify objects in the sky that vary in brightness from simulated LSST time-series data, with all its challenges of non-representativity? In this note we explain the need for a data challenge to help classify such astronomical sources and describe the PLAsTiCC data set and Kaggle data challenge, noting that while the references are provided for context, they are not needed to participate in the challenge.

astro-ph.IM↗

arXiv2018

Multi-Messenger Astrophysics: Harnessing the Data Revolution

The past year has witnessed discovery of the first identified counterparts to a gravitational wave transient (GW 170817A) and a very high-energy neutrino (IceCube-170922A). These source identifications, and ensuing detailed studies, have realized longstanding dreams of astronomers and physicists to routinely carry out observations of cosmic sources by other than electromagnetic means, and inaugurated the era of "multi-messenger" astronomy. While this new era promises extraordinary physical insights into the universe, it brings with it new challenges, including: highly heterogeneous, high-volume, high-velocity datasets; globe-spanning cross-disciplinary teams of researchers, regularly brought together into transient collaborations; an extraordinary breadth and depth of domain-specific knowledge and computing resources required to anticipate, model, and interpret observations; and the routine need for adaptive, distributed, rapid-response observing campaigns to fully exploit the scientific potential of each source. We argue, therefore, that the time is ripe for the community to conceive and propose an Institute for Multi-Messenger Astrophysics that would coordinate its resources in a sustained and strategic fashion to efficiently address these challenges, while simultaneously serving as a center for education and key supporting activities. In this fashion, we can prepare now to realize the bright future that we see, beyond, through these newly opened windows onto the universe.

astro-ph.IM↗

arXiv2016

A cosmographic analysis of the transition to acceleration using SN-Ia and BAO

We explore the distance-redshift relation using a cosmographic methodology, and show how the cosmographic parameters can be used to determine the redshift of transition from deceleration to acceleration. Such a transition at a low redshift occupies only a small region of the available parameter space, and the prior assumption of an early period of deceleration can significantly change the posterior constraints. We use available type Ia Supernovae (SN-Ia) and Baryon Acoustic Oscillation (BAO) data sets to determine the cosmographic deceleration $q_0$, jerk $j_0$, snap $s_0$ and lerk $l_0$ parameters. The parameters are consistent with the $Λ$CDM model for a flat universe within 2-sigma. We derive constraints on the redshift of transition from deceleration to acceleration for the different expansions, and find $z_{\rm acc} > 0.14$ at 95% confidence in the most conservative case.

astro-ph.CO↗