Active deep learning method for the discovery of objects of interest in large spectroscopic surveys

Abstract

Current archives of the LAMOST telescope contain millions of pipeline-processed spectra that have probably never been seen by human eyes. Most of the rare objects with interesting physical properties, however, can only be identified by visual analysis of their characteristic spectral features. A proper combination of interactive visualisation with modern machine learning techniques opens new ways to discover such objects. We apply active learning classification supported by deep convolutional networks to automatically identify complex emission-line shapes in multi-million spectra archives. We used the pool-based uncertainty sampling active learning driven by a custom-designed deep convolutional neural network with 12 layers inspired by VGGNet, AlexNet, and ZFNet, but adapted for one-dimensional feature vectors. The unlabelled pool set is represented by 4.1 million spectra from the LAMOST DR2 survey. The initial training of the network was performed on a labelled set of about 13000 spectra obtained in the region around H$\alpha$ by the 2m Perek telescope of the Ondřejov observatory, which mostly contains spectra of Be and related early-type stars. The differences between the Ondřejov intermediate-resolution and the LAMOST low-resolution spectrographs were compensated for by Gaussian blurring. After several iterations, the network was able to successfully identify emission-line stars with an error smaller than 6.5%. Using the technology of the Virtual Observatory to visualise the results, we discovered 1013 spectra of 948 new candidates of emission-line objects in addition to 664 spectra of 549 objects that are listed in SIMBAD and 2644 spectra of 2291 objects identified in an earlier paper of a Chinese group led by Wen Hou. The most interesting objects with unusual spectral properties are discussed in detail.

arXiv commentTo appear in Astronomy & Astrophysics Section 14- Catalogs and Data, 15 pages, 13 figures

References in corpus (27)

• SMOTE: Synthetic Minority Over-sampling Technique
• The First Data Release (DR1) of the LAMOST general survey
• The SEGUE Stellar Parameter Pipeline. I. Description and Initial Validation Tests
• Improving galaxy morphologies for SDSS with Deep Learning
• Deep Neural Networks to Enable Real-time Multimessenger Astrophysics
• Fifty Years of Pulsar Candidate Selection: From simple filters to a new principled real-time classification approach
• On the formation of H-alpha line emission around classical T Tauri stars
• Automatic Determination of Stellar Atmospheric Parameters and Construction of Stellar Spectral Templates of the Guoshoujing Telescope (LAMOST)
• Classifying Radio Galaxies with Convolutional Neural Network
• M31 Transverse Velocity and Local Group Mass from Satellite Kinematics
• Optimizing spectroscopic follow-up strategies for supernova photometric classification with active learning
• The First Hypervelocity Star from the LAMOST Survey
• Active Learning to Overcome Sample Selection Bias: Application to Photometric Variable Star Classification
• Kinematical structure of the circumstellar environments of galactic B[e]-type stars
• Hierarchical progressive surveys. Multi-resolution HEALPix data structures for astronomical images, catalogues, and 3-dimensional data cubes
• Radio Galaxy Zoo: Machine learning for radio source host galaxy cross-identification
• The FIRST Classifier: Compact and Extended Radio Galaxy Classification using Deep Convolutional Neural Networks
• Tool for Automatic Measurement of Equivalent width (TAME)
• Flux rope proxies and fan-spine structures in active region NOAA 11897
• ROBOSPECT: Automated Equivalent Width Measurement
• Spectroscopic Analysis in the Virtual Observatory Environment with SPLAT-VO
• Virtual Observatory Publishing with DaCHS
• SAMP, the Simple Application Messaging Protocol: Letting applications talk to each other
• The Gaia-ESO Survey: Catalogue of H$\alpha$ emission stars
• An Active Instance-based Machine Learning method for Stellar Population Studies
• A General Approach to Domain Adaptation with Applications in Astronomy
• Identification of Interesting Objects in Large Spectral Surveys Using Highly Parallelized Machine Learning

Cited by in corpus (11)

• Data mining techniques on astronomical spectra data. II : Classification Analysis
• SUPPNet: Neural network for stellar spectrum normalisation
• A method for finding anomalous astronomical light curves and their analogs
• A catalog of early-type Hα emission line stars and 58 newly confirmed Herbig Ae/Bes from LAMOST DR7
• A Machine Learning Approach to Integral Field Unit Spectroscopy Observations: II. HII Region LineRatios
• Investigation of stellar magnetic activity using variational autoencoder based on low-resolution spectroscopic survey
• Estimation of Stellar Parameters and Mass Accretion Rate of Classical T Tauri Stars from LAMOST DR6
• Unveiling compact planetary nebulae: Broad-band survey analysis and LAMOST confirmation
• LAMOST J040901.83+329355.6 -- a new Galactic star with Wolf--Rayet characteristics on a post-AGB to CSPN transitional stage
• A Catalog of 45,206 Hα Emission-line Stars from LAMOST MRS DR7
• The identification of new Herbig Ae/Be stars from LAMOST DR7