ExoMiner: A Highly Accurate and Explainable Deep Learning Classifier that Validates 301 New Exoplanets

Abstract

The kepler and TESS missions have generated over 100,000 potential transit signals that must be processed in order to create a catalog of planet candidates. During the last few years, there has been a growing interest in using machine learning to analyze these data in search of new exoplanets. Different from the existing machine learning works, ExoMiner, the proposed deep learning classifier in this work, mimics how domain experts examine diagnostic tests to vet a transit signal. ExoMiner is a highly accurate, explainable, and robust classifier that 1) allows us to validate 301 new exoplanets from the MAST Kepler Archive and 2) is general enough to be applied across missions such as the on-going TESS mission. We perform an extensive experimental study to verify that ExoMiner is more reliable and accurate than the existing transit signal classifiers in terms of different classification and ranking metrics. For example, for a fixed precision value of 99%, ExoMiner retrieves 93.6% of all exoplanets in the test set (i.e., recall=0.936) while this rate is 76.3% for the best existing classifier. Furthermore, the modular design of ExoMiner favors its explainability. We introduce a simple explainability framework that provides experts with feedback on why ExoMiner classifies a transit signal into a specific class label (e.g., planet candidate or not planet candidate).

arXiv commentAccepted for Publication in Astrophysical Journals, November 12, 2021

References in corpus (40)

• Gaia Data Release 2. Summary of the contents and survey properties
• Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization
• The Transiting Exoplanet Survey Satellite
• On Calibration of Modern Neural Networks
• The K2 Mission: Characterization and Early results
• Characteristics of planetary candidates observed by Kepler, II: Analysis of the first four months of data
• Kepler Input Catalog: Photometric Calibration and Stellar Classification
• Overview of the Kepler Science Processing Pipeline
• Star counts in the Galaxy. Simulating from very deep to very shallow photometric surveys with the TRILEGAL code
• False positive probabilties for all Kepler Objects of Interest: 1284 newly validated planets and 428 likely false positives
• Planetary Candidates Observed by Kepler. VIII. A Fully Automated Catalog With Measured Completeness and Reliability Based on Data Release 25
• Deep Learning is Robust to Massive Label Noise
• The TESS Objects of Interest Catalog from the TESS Prime Mission
• Identifying Exoplanets with Deep Learning: A Five Planet Resonant Chain around Kepler-80 and an Eighth Planet around Kepler-90
• Kepler Data Validation I -- Architecture, Diagnostic Tests, and Data Products for Vetting Transiting Planet Candidates
• SOPHIE velocimetry of Kepler transit candidates XVII. The physical properties of giant exoplanets within 400 days of period
• An Efficient Automated Validation Procedure for Exoplanet Transit Candidates
• Kepler Data Validation II -- Transit Model Fitting and Multiple-Planet Search
• On the Low False Positive Probabilities of Kepler Planet Candidates
• Planetary Candidates Observed by Kepler. VII. The First Fully Uniform Catalog Based on The Entire 48 Month Dataset (Q1-Q17 DR24)
• Planet Hunters: A Transiting Circumbinary Planet in a Quadruple Star System
• Identification of Background False Positives from Kepler Data
• Transiting extrasolar planetary candidates in the Galactic bulge
• Testing blend scenarios for extrasolar transiting planet candidates. I. -- OGLE-TR-33: a false positive
• Contamination in the Kepler Field. Identification of 685 KOIs as False Positives Via Ephemeris Matching Based On Q1-Q12 Data
• Discovery and Rossiter-McLaughlin Effect of Exoplanet Kepler-8b
• Automatic Classification of Kepler Planetary Transit Candidates
• Detection of Potential Transit Signals in 17 Quarters of Kepler Data: Results of the Final Kepler Mission Transiting Planet Search (DR25)
• Scientific Domain Knowledge Improves Exoplanet Transit Classification with Deep Learning
• Identifying Exoplanets with Deep Learning III: Automated Triage and Vetting of TESS Candidates
• Identifying Exoplanets with Deep Learning II: Two New Super-Earths Uncovered by a Neural Network in K2 Data
• Planet Hunters TESS I: TOI 813, a subgiant hosting a transiting Saturn-sized planet on an 84-day orbit
• Rapid Classification of TESS Planet Candidates with Convolutional Neural Networks
• Transit Shapes and Self Organising Maps as a Tool for Ranking Planetary Candidates: Application to Kepler and K2
• Detection of Potential Transit Signals in 17 Quarters of Kepler Mission Data
• Exoplanet Validation with Machine Learning: 50 new validated Kepler planets
• Planets, candidates, and binaries from the CoRoT/Exoplanet programme: the CoRoT transit catalogue
• Classifying Exoplanet Candidates with Convolutional Neural Networks: Application to the Next Generation Transit Survey
• Re-Evaluating Small Long-Period Confirmed Planets From Kepler
• Following up the Kepler field: Masses of Targets for transit timing and atmospheric characterization

Cited by in corpus (21)

• The Kepler Giant Planet Search. I: A Decade of Kepler Planet-host Radial Velocities from W. M. Keck Observatory
• The Ariel Target List: The Impact of TESS and the Potential for Characterising Multiple Planets Within a System
• ExoMDN: Rapid characterization of exoplanet interior structures with Mixture Density Networks
• The SHERLOCK pipeline: new exoplanet candidates in the WASP-16, HAT-P-27, HAT-P-26, and TOI-2411 systems
• Multiplicity Boost Of Transit Signal Classifiers: Validation of 69 New Exoplanets Using The Multiplicity Boost of ExoMiner
• Dynamical Evolution of Closely Packed Multiple Planetary Systems Subject to Atmospheric Mass-Loss
• Identifying Light-curve Signals with a Deep Learning Based Object Detection Algorithm. II. A General Light Curve Classification Framework
• The TESS Triple-9 Catalog II: a new set of 999 uniformly-vetted exoplanet candidates
• Insightful analysis of historical sources at scales beyond human capabilities using unsupervised Machine Learning and XAI
• exoMMR: a New Python Package to Confirm and Characterize Mean Motion Resonances
• Exoplanet Transit Candidate Identification in TESS Full-Frame Images via a Transformer-Based Algorithm
• Datacube segmentation via Deep Spectral Clustering
• DART-Vetter: A Deep LeARning Tool for automatic triage of exoplanet candidates
• TelescopeML -- I. An End-to-End Python Package for Interpreting Telescope Datasets through Training Machine Learning Models, Generating Statistical Reports, and Visualizing Results
• Transit Hunt for Young and Maturing Exoplanets (THYME) VIII: a Pleiades-age association harboring two transiting planetary systems from Kepler
• Exploring Long-period Architectures: Four New Planet Candidates from Kepler with Periods >342 days
• DeepRed: an architecture for redshift estimation
• Assessing Galaxy Rotation Kinematics: Insights from Convolutional Neural Networks on Velocity Variations
• Panopticon: a novel deep learning model to detect single transit events with no prior data filtering in PLATO light curves
• Exoplanet transit search at the detection limit: detection and false alarm vetting pipeline
• ExoDNN: Boosting exoplanet detection with artificial intelligence. Application to Gaia Data Release 3