Active Anomaly Detection for time-domain discoveries
arXiv:1909.13260 · doi:10.1051/0004-6361/202037709
Abstract
We present the first evidence that adaptive learning techniques can boost the discovery of unusual objects within astronomical light curve data sets. Our method follows an active learning strategy where the learning algorithm chooses objects which can potentially improve the learner if additional information about them is provided. This new information is subsequently used to update the machine learning model, allowing its accuracy to evolve with each new information. For the case of anomaly detection, the algorithm aims to maximize the number of scientifically interesting anomalies presented to the expert by slightly modifying the weights of a traditional Isolation Forest (IF) at each iteration. In order to demonstrate the potential of such techniques, we apply the Active Anomaly Discovery (AAD) algorithm to 2 data sets: simulated light curves from the PLAsTiCC challenge and real light curves from the Open Supernova Catalog. We compare the AAD results to those of a static IF. For both methods, we performed a detailed analysis for all objects with the ~2% highest anomaly scores. We show that, in the real data scenario, AAD was able to identify ~80\% more true anomalies than the IF. This result is the first evidence that AAD algorithms can play a central role in the search for new physics in the era of large scale sky surveys.
arXiv comment10 pages, 5 figures, updated to include PLAsTiCC results
References in corpus (39)
- Scikit-learn: Machine Learning in Python
- The NumPy array: a structure for efficient numerical computation
- Luminous Supernovae
- An Open Catalog for Supernova Data
- Type Iax Supernovae: A New Class of Stellar Explosion
- The Carnegie Supernova Project: First Photometry Data Release of Low-Redshift Type Ia Supernovae
- The Sloan Digital Sky Survey Monitor Telescope Pipeline
- The Observer's Guide to the Gamma-Ray Burst-Supernova Connection
- The Carnegie Supernova Project: Second Photometry Data Release of Low-Redshift Type Ia Supernovae
- The Spectroscopic Diversity of Type Ia Supernovae
- Massive stars exploding in a He-rich circumstellar medium. I. Type Ibn (SN 2006jc-like) events
- Super-luminous supernovae at redshifts of 2.05 and 3.90
- Galaxy Zoo: Probabilistic Morphology through Bayesian CNNs and Active Learning
- Spectroscopy of Type Ia Supernovae by the Carnegie Supernova Project
- The Foundation Supernova Survey: Motivation, Design, Implementation, and First Data Release
- The Extremes of Thermonuclear Supernovae
- Superluminous supernovae
- Models and Simulations for the Photometric LSST Astronomical Time Series Classification Challenge (PLAsTiCC)
- The weirdest SDSS galaxies: results from an outlier detection algorithm
- Super Luminous Supernovae as standardizable candles and high redshift distance probes
- Optimizing spectroscopic follow-up strategies for supernova photometric classification with active learning
- Anomaly detection for machine learning redshifts applied to SDSS galaxies
- Active Learning to Overcome Sample Selection Bias: Application to Photometric Variable Star Classification
- PS1-12sk is a Peculiar Supernova From a He-rich Progenitor System in a Brightest Cluster Galaxy Environment
- Anomaly detection in the Zwicky Transient Facility DR3
- The fast declining Type Ia supernova 2003gs, and evidence for a significant dispersion in near-infrared absolute magnitudes of fast decliners at maximum light
- Systematic Serendipity: A Test of Unsupervised Machine Learning as a Method for Anomaly Detection
- Supervised detection of anomalous light-curves in massive astronomical catalogs
- Space Telescope and Optical Reverberation Mapping Project. XII. Broad-Line Region Modeling of NGC 5548
- NTT and NOT spectroscopy of SDSS-II supernovae
- Progenitors of low-luminosity Type II-Plateau supernovae
- Anomaly Detection in the Open Supernova Catalog
- Incorporating Feedback into Tree-based Anomaly Detection
- Massive stars exploding in a He-rich circumstellar medium. VI. Observations of two distant Type Ibn supernova candidates discovered by La Silla-QUEST
- The Photometric LSST Astronomical Time-series Classification Challenge (PLAsTiCC): Data set
- Defining Photometric Peculiar Type Ia Supernovae
- An Active Instance-based Machine Learning method for Stellar Population Studies
- Active Anomaly Detection via Ensembles
- Peculiar Supernovae
Cited by in corpus (28)
- Fink, a new generation of broker for the LSST community
- Anomaly detection in the Zwicky Transient Facility DR3
- A Deep Learning Approach for Active Anomaly Detection of Extragalactic Transients
- Searching for changing-state AGNs in massive datasets -- I: applying deep learning and anomaly detection techniques to find AGNs with anomalous variability behaviours
- ParSNIP: Generative Models of Transient Light Curves with Physics-Enabled Deep Learning
- The SNAD Viewer: Everything You Want to Know about Your Favorite ZTF Object
- Fink: early supernovae Ia classification using active learning
- Light curve classification with recurrent neural networks for GOTO: dealing with imbalanced data
- Applications of Machine Learning to Predicting Core-collapse Supernova Explosion Outcomes
- Real-Time Detection of Anomalies in Large-Scale Transient Surveys
- A method for finding anomalous astronomical light curves and their analogs
- Inferencing Progenitor and Explosion Properties of Evolving Core-collapse Supernovae from Zwicky Transient Facility Light Curves
- Preparing to discover the unknown with Rubin LSST -- I: Time domain
- ATAT: Astronomical Transformer for time series And Tabular data
- Identifying outliers in astronomical images with unsupervised machine learning
- Bursts from Space: MeerKAT - The first citizen science project dedicated to commensal radio transients
- Supernova search with active learning in ZTF DR3
- SNAD catalogue of M-dwarf flares from the Zwicky Transient Facility
- SNAD Transient Miner: Finding Missed Transient Events in ZTF DR4 using k-D trees
- Understanding of the properties of neural network approaches for transient light curve approximations
- A Classifier-Based Approach to Multi-Class Anomaly Detection for Astronomical Transients
- A Data-Driven Search For Mid-Infrared Excesses Among Five Million Main-Sequence FGK Stars
- Finding radio transients with anomaly detection and active learning based on volunteer classifications
- Real-Time Active Learning for optimised spectroscopic follow-up: Enhancing early SN Ia classification with the Fink broker
- Real-bogus scores for active anomaly detection
- Exploring the Universe with SNAD: Anomaly Detection in Astronomy
- Identification of Anomalous E+A Galaxies in GAMA Using an Isolation Forest
- Anomaly Detection for Multivariate Time Series of Exotic Supernovae