Bulletin de la Société Royale des Sciences de Liège Bulletin de la Société Royale des Sciences de Liège -  Volume 88 - Année 2019  Actes de colloques  2nd Belgo-Indian Network for Astronomy & Astrophysics (BINA) workshop - October 2018 - Brussels, Belgium 

Detecting Outliers in SDSS using Convolutional Neural Network

Kaushal Sharma
Inter University Centre for Astronomy and Astrophysics (IUCAA), Pune, India,  kaushals@iucaa.in
Ajit Kembhavi
Inter University Centre for Astronomy and Astrophysics (IUCAA), Pune, India
Aniruddha Kembhavi
Allen Institute for Artificial Intelligence (AI2), Seattle, USA
Thirupathi Sivarani
Indian Institute of Astrophysics (IIA), Bengaluru, India
Sheelu Abraham
Inter University Centre for Astronomy and Astrophysics (IUCAA), Pune, India

Abstract

We propose an automated algorithm based on Convolutional Neural Network (CNN) for the detection of peculiar objects in large databases using their spectral observations. A convolutional neural network is a class of deep-learning algorithms which allows the detection of significant features/patterns in sequential data like images, audio, time-series etc. by applying convolutional neurons (kernels) along the sequence. For detecting unusual spectra, we use eight-layer deep convolutional network with autoencoder architecture on ~ 60,000 spectra collected from the Sloan Digital Sky Survey. The training of the network is done in an unsupervised manner. We show that the trained network is able to retrieve the spectra of rare objects from a large collection of spectra. Such algorithms can easily be rescaled to other surveys and therefore can serve as a potential component of the data reduction pipelines for automatically detecting spectra with unusual features and recovering defective spectra.

Keywords : astronomical databases, spectroscopy, deep learning, autoencoder, outlier detection

Pour citer cet article

Kaushal Sharma, Ajit Kembhavi, Aniruddha Kembhavi, Thirupathi Sivarani & Sheelu Abraham, «Detecting Outliers in SDSS using Convolutional Neural Network», Bulletin de la Société Royale des Sciences de Liège [En ligne], Volume 88 - Année 2019, Actes de colloques, 2nd Belgo-Indian Network for Astronomy & Astrophysics (BINA) workshop - October 2018 - Brussels, Belgium, 174-181 URL : https://popups.uliege.be/0037-9565/index.php?id=8811.