Classification of Periodic Variable Stars with Novel Cyclic-Permutation Invariant Neural Networks

Abstract

Neural networks (NNs) have been shown to be competitive against state-of-the-art feature engineering and random forest (RF) classification of periodic variable stars. Although previous work utilising NNs has made use of periodicity by period folding multiple-cycle time-series into a single cycle -- from time-space to phase-space -- no approach to date has taken advantage of the fact that network predictions should be invariant to the initial phase of the period-folded sequence. Initial phase is exogenous to the physical origin of the variability and should thus be factored out. Here, we present cyclic-permutation invariant networks, a novel class of NNs for which invariance to phase shifts is guaranteed through polar coordinate convolutions, which we implement by means of "Symmetry Padding." Across three different datasets of variable star light curves, we show that two implementations of the cyclic-permutation invariant network: the iTCN and the iResNet, consistently outperform non-invariant baselines and reduce overall error rates by between 4% to 22%. Over a 10-class OGLE-III sample, the iTCN/iResNet achieves an average per-class accuracy of 93.4%/93.3%, compared to RNN/RF accuracies of 70.5%/89.5% in a recent study using the same data. Finding improvement on a non-astronomy benchmark, we suggest that the methodology introduced here should also be applicable to a wide range of science domains where periodic data abounds due to physical symmetries.

arXiv commentAccepted to MNRAS. Extended abstract accepted to NeurIPS 2020: Machine Learning for the Physical Sciences. Initially appeared in ICLR 2020: Fundamental Science in the Era of AI. Open source implementation at https://github.com/kmzzhang/periodicnetwork. 8 pages, 4 figures

References in corpus (37)

• Adam: A Method for Stochastic Optimization
• An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling
• Machine learning and the physical sciences
• VISTA Variables in the Via Lactea (VVV): The public ESO near-IR variability survey of the Milky Way
• A Simple Way to Initialize Recurrent Networks of Rectified Linear Units
• Absolute dimensions of the M-type eclipsing binary YY Geminorum (Castor C): a challenge to evolutionary models in the lower main-sequence
• The ASAS-SN Catalog of Variable Stars I: The Serendipitous Survey
• Identifying Exoplanets with Deep Learning: A Five Planet Resonant Chain around Kepler-80 and an Eighth Planet around Kepler-90
• On Machine-Learned Classification of Variable Stars with Sparse and Noisy Time-Series Data
• Unitary Evolution Recurrent Neural Networks
• Star-galaxy Classification Using Deep Convolutional Neural Networks
• The ASAS-SN Catalog of Variable Stars II: Uniform Classification of 412,000 Known Variables
• A three-dimensional map of the Milky Way using classical Cepheid variable stars
• SuperNNova: an open-source framework for Bayesian, Neural Network based supernova classification
• Zoneout: Regularizing RNNs by Randomly Preserving Hidden Activations
• FETCH: A deep-learning based classifier for fast transient classification
• The First DIRECT Distance Determination to a Detached Eclipsing Binary in M33
• Gaia Data Release 2: Variable stars in the colour-absolute magnitude diagram
• The Extragalactic Distance Scale
• A recurrent neural network for classification of unevenly sampled variable stars
• QCD-Aware Recursive Neural Networks for Jet Physics
• Machine Learning-based Brokers for Real-time Classification of the LSST Alert Stream
• A Package for the Automated Classification of Periodic Variable Stars
• Eclipsing Binaries as Astrophysical Laboratories: Internal Structure, Convective Core Overshooting and Evolution of the B-star Components of V380 Cygni
• Spatial Transformer Networks
• Physical Symmetries Embedded in Neural Networks
• Scalable End-to-end Recurrent Neural Network for Variable star classification
• A delta Scuti distance to the Large Magellanic Cloud
• Deep multi-survey classification of variable stars
• On Neural Architectures for Astronomical Time-series Classification with Application to Variable Stars
• deepCR: Cosmic Ray Rejection with Deep Learning
• Deep modeling of quasar variability
• The Extinction Toward the Galactic Bulge from RR Lyrae Stars
• Into the Darkness: Classical and Type II Cepheids in the Zona Galactica Incognita
• Deep Neural Network Classifier for Variable Stars with Novelty Detection Capability
• FATS: Feature Analysis for Time Series
• Near-Infrared Search for Fundamental-mode RR Lyrae Stars Toward the Inner Bulge by Deep Learning

Cited by in corpus (12)

• Classifying Kepler light curves for 12,000 A and F stars using supervised feature-based machine learning
• Real-Time Likelihood-Free Inference of Roman Binary Microlensing Events with Amortized Neural Posterior Estimation
• Deep Attention-Based Supernovae Classification of Multi-Band Light-Curves
• Minute-cadence Observations of the LAMOST Fields with the TMTS: I. Methodology of Detecting Short-period Variables and Results from the first-year Survey
• Variable star classification with a Multiple-Input Neural Network
• Searching for Anomalies in the ZTF Catalog of Periodic Variable Stars
• Application of Convolutional Neural Networks to time domain astrophysics. 2D image analysis of OGLE light curves
• Heartbeat Stars Recognition Based on Recurrent Neural Networks: Method and Validation
• Informative regularization for a multi-layer perceptron RR Lyrae classifier under data shift
• Sparse logistic regression for RR Lyrae vs binaries classification
• Light curve fingerprints: an automated approach to the extraction of X-ray variability patterns with feature aggregation -- an example application to GRS 1915+105
• Pushing the Limit of Asteroseismic Detection for Cool Dwarfs using TESS and Deep Learning