Caltech research presents “SNIascore”, a method for spectroscopic classification of thermonuclear supernovae (SNe Ia) based on very low resolution (R 100) data based on deep learning. The goal of SNIascore is to fully automate the classification of SNe Ia with a very low false positive rate (FPR) so that people do less work.
In the public Zwicky Transient Facility (ZTF), Bright Transient Surveys and other large-scale SN (BTS) classification efforts. They use a recurrent long-term memory neural network architecture that works in both directions and recurrent unit layers controlled by gates. SNIascore has an FPR of 0.6% and can classify up to 90% of the low-resolution SN Ia spectra of the BTS. SNIascore performs binary classification and regression to predict the redshifts of safe SNe Ia. Using SNIascore with the magnitude-limited ZTF-BTS survey (70% SNe Ia) reduces the number of spectra that must be classified or confirmed by an individual by approximately 60%.
In addition, SNIascore enables SN Ia classifications to be automatically announced to the public in real-time immediately after a nighttime observation.
Astronomers attempt to answer some of the most exciting and important scientific questions of our time, which often requires them to gather lots of information about various cosmic events. So modern astronomical observatories have turned into machines that send astronomers tens of thousands of alerts and images every night. This is especially true in time-domain astronomy, where scientists look for objects that change rapidly, called transients. These include stars that explode and die, called supernovae, and black holes that eat orbiting stars, asteroids, and other things.
The proposed machine learning algorithm is much faster at classifying possible supernova candidates and shares the results with the astronomical community. It also gives astronomers more time to work on other scientific questions. With SNIascore, the process takes about ten minutes instead of two to three days. Explosions in space must be found as soon as possible so scientists can learn more about how they work.
Currently, SNIascore can only classify Type Ia supernovae, which astronomers use as “standard candles” to measure how fast the Universe is expanding. These stars die, and when they explode, they do so with a powerful thermonuclear bang.
SNIascore is now set up to work with the SEDM (Spectral Energy Distribution Machine) spectrograph, located in a dome of the Palomar Observatory, just a few hundred meters from the ZTF camera. Constantly looking up at the sky, the ZTF sends tens of thousands of alerts every night to astronomers around the world about possible cosmic transients. The SEDM spectrograph sets out to track and observe the most interesting ones. It makes a spectrum of the cosmic event that shows how strong different frequencies of light the telescopic camera caught were. Astronomers can be sure of what type of event they are seeing because of this spectrum. The researcher used clever machine learning techniques to teach SNIascore to read the SEDM spectra well.
Researchers are currently making modifications to SNIascore to work with the new SEDMv2 spectrograph that will be installed on the 2.1m telescope. SEDMv2 will be an improved version of SEDM. It will be able to find and classify dimmer supernovae. At the moment, SNIascore classifies about two supernovae every night on average. This number could double if SEDMv2 is used.
The benefits of SNIascore go beyond the fast and reliable creation of large datasets of supernovae. Astronomers looking for other types of transient events can quickly rule out candidates that SNIascore says are supernovae. This means no telescope time is wasted following them in search of other types of explosions in space.
Other attempts to classify transient events also use machine learning, but they only use the event’s “light curve,” or the amount of light the telescope sees over time. SNIascore is good because it is trained on and uses spectroscopic information, which is the only reliable way to confirm what most transients are. The code for the algorithm is public so other groups can modify it to make it work with their telescopes.
Try this paper and reference article. All credit for this research goes to the researchers on this project. Also don’t forget to participate our Reddit page and Discord Channelwhere we share the latest AI research news, cool AI projects and more.
Ashish Kumar is a consulting intern at MarktechPost. He is currently pursuing his Btech studies at Indian Institute of Technology (IIT), Kanpur. He is passionate about exploring new technological advances and their application in real life.
