Classifying Exoplanet Candidates with Convolutional Neural Networks: Application to the Next Generation Transit Survey

Abstract

Vetting of exoplanet candidates in transit surveys is a manual process, which suffers from a large number of false positives and a lack of consistency. Previous work has shown that Convolutional Neural Networks (CNN) provide an efficient solution to these problems. Here, we apply a CNN to classify planet candidates from the Next Generation Transit Survey (NGTS). For training datasets we compare both real data with injected planetary transits and fully-simulated data, as well as how their different compositions affect network performance. We show that fewer hand labelled lightcurves can be utilised, while still achieving competitive results. With our best model, we achieve an AUC (area under the curve) score of $(95.6\pm 0.2)\%$ and an accuracy of $(88.5\pm 0.3)\%$ on our unseen test data, as well as $(76.5\pm 0.4)\%$ and $(74.6\pm 1.1)\%$ in comparison to our existing manual classifications. The neural network recovers 13 out of 14 confirmed planets observed by NGTS, with high probability. We use simulated data to show that the overall network performance is resilient to mislabelling of the training dataset, a problem that might arise due to unidentified, low signal-to-noise transits. Using a CNN, the time required for vetting can be reduced by half, while still recovering the vast majority of manually flagged candidates. In addition, we identify many new candidates with high probabilities which were not flagged by human vetters.

21 pages, 10 figures, Accepted for publication in MNRAS

References in corpus (49)

• Adam: A Method for Stochastic Optimization
• Improving neural networks by preventing co-adaptation of feature detectors
• The Transiting Exoplanet Survey Satellite
• The WASP Project and the SuperWASP Cameras
• The NASA Exoplanet Archive: Data and Tools for Exoplanet Research
• Overview of the Kepler Science Processing Pipeline
• The effect of red noise on planetary transit detection
• Identifying Exoplanets with Deep Learning: A Five Planet Resonant Chain around Kepler-80 and an Eighth Planet around Kepler-90
• Efficient identification of exoplanetary transit candidates from SuperWASP light curves
• Revisiting Unreasonable Effectiveness of Data in Deep Learning Era
• KELT-1b: A Strongly Irradiated, Highly Inflated, Short Period, 27 Jupiter-mass Companion Transiting a mid-F Star
• Kepler Flares II: The Temporal Morphology of White-Light Flares on GJ 1243
• The Next Generation Transit Survey (NGTS)
• SOPHIE velocimetry of Kepler transit candidates XVII. The physical properties of giant exoplanets within 400 days of period
• HAT-P-1b: A Large-Radius, Low-Density Exoplanet Transiting one Member of a Stellar Binary
• A fast hybrid algorithm for exoplanetary transit searches
• Learning Deep Features for Discriminative Localization
• Probabilistic Random Forest: A machine learning algorithm for noisy datasets
• The XO Project: Searching for Transiting Extra-solar Planet Candidates
• ELLC - a fast, flexible light curve model for detached eclipsing binary stars and transiting exoplanets
• Gradient Descent with Early Stopping is Provably Robust to Label Noise for Overparameterized Neural Networks
• Transiting hot Jupiters from WASP-South, Euler and TRAPPIST: WASP-95b to WASP-101b
• Searching for Exoplanets Using Artificial Intelligence
• Automatic Classification of Kepler Planetary Transit Candidates
• NGTS-1b: A hot Jupiter transiting an M-dwarf
• WASP-South transiting exoplanets: WASP-130b, WASP-131b, WASP-132b, WASP-139b, WASP-140b, WASP-141b & WASP-142b
• NGTS-4b: A sub-Neptune Transiting in the Desert
• 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
• Machine-learning Approaches to Exoplanet Transit Detection and Candidate Validation in Wide-field Ground-based Surveys
• A Machine Learning Technique to Identify Transit Shaped Signals
• Shallow Transits - Deep Learning I: Feasibility Study of Deep Learning to Detect Periodic Transits of Exoplanets
• A study of the performance of the transit detection tool DST in space-based surveys. Application of the CoRoT pipeline to Kepler data
• KELT-10b: The First Transiting Exoplanet from the KELT-South Survey -- A Hot Sub-Jupiter Transiting a V = 10.7 Early G-Star
• MASCARA-1 b: A hot Jupiter transiting a bright $m_V=8.3$ A-star in a misaligned orbit
• Transit Shapes and Self Organising Maps as a Tool for Ranking Planetary Candidates: Application to Kepler and K2
• Planets, candidates, and binaries from the CoRoT/Exoplanet programme: the CoRoT transit catalogue
• Centroid vetting of transiting planet candidates from the Next Generation Transit Survey
• SIDRA: a blind algorithm for signal detection in photometric surveys
• Automatic vetting of planet candidates from ground based surveys: Machine learning with NGTS
• Unmasking the hidden NGTS-3Ab: a hot Jupiter in an unresolved binary system
• NGTS-6b: An Ultra Short Period Hot-Jupiter Orbiting an Old K Dwarf
• NGTS-2b: An inflated hot-Jupiter transiting a bright F-dwarf
• HAT-P-42b and HAT-P-43b. Two Inflated Transiting Hot Jupiters from the HATNet Survey
• A New Yield Simulator for Transiting Planets and False Positives: Application to the Next Generation Transit Survey
• NGTS-5b: a highly inflated planet offering insights into the sub-Jovian desert
• The PDS 110 observing campaign - photometric and spectroscopic observations reveal eclipses are aperiodic
• Transdet: a matched-filter based algorithm for transit detection - application to simulated COROT light curves

Cited by in corpus (17)

• ExoMiner: A Highly Accurate and Explainable Deep Learning Classifier that Validates 301 New Exoplanets
• Machine learning inference of the interior structure of low-mass exoplanets
• Exoplanet Validation with Machine Learning: 50 new validated Kepler planets
• ExoMDN: Rapid characterization of exoplanet interior structures with Mixture Density Networks
• NGTS-19b : A high mass transiting brown dwarf in a 17-day eccentric orbit
• Identifying Exoplanets with Deep Learning. V. Improved Light Curve Classification for TESS Full Frame Image Observations
• NGTS-14Ab: a Neptune-sized transiting planet in the desert
• Identify Light-Curve Signals with Deep Learning Based Object Detection Algorithm. I. Transit Detection
• NGTS-12b: A sub-Saturn mass transiting exoplanet in a 7.53 day orbit
• NGTS 15b, 16b, 17b and 18b: four hot Jupiters from the Next Generation Transit Survey
• One-dimensional Convolutional Neural Networks for Detecting Transiting Exoplanets
• Identifying Exoplanets with Deep Learning. IV. Removing Stellar Activity Signals from Radial Velocity Measurements Using Neural Networks
• The GPU Phase Folding and Deep Learning Method for Detecting Exoplanet Transits
• Computing Transiting Exoplanet Parameters with 1D Convolutional Neural Networks
• DART-Vetter: A Deep LeARning Tool for automatic triage of exoplanet candidates
• Photometric Search for Exomoons by using Convolutional Neural Networks
• A one-armed CNN for exoplanet detection from light curves