Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userIdentifying Exoplanets with Deep Learning III: Automated Triage and Vetting of TESS Candidates
120
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
NASAs Transiting Exoplanet Survey Satellite (TESS) presents us with an unprecedented volume of space-based photometric observations that must be analyzed in an efficient and unbiased manner. With at least $sim1,000,000$ new light curves generated every month from full frame images alone, automated planet candidate identification has become an attractive alternative to human vetting. Here we present a deep learning model capable of performing triage and vetting on TESS candidates. Our model is modified from an existing neural network designed to automatically classify Kepler candidates, and is the first neural network to be trained and tested on real TESS data. In triage mode, our model can distinguish transit-like signals (planet candidates and eclipsing binaries) from stellar variability and instrumental noise with an average precision (the weighted mean of precisions over all classification thresholds) of 97.0% and an accuracy of 97.4%. In vetting mode, the model is trained to identify only planet candidates with the help of newly added scientific domain knowledge, and achieves an average precision of 69.3% and an accuracy of 97.8%. We apply our model on new data from Sector 6, and present 288 new signals that received the highest scores in triage and vetting and were also identified as planet candidates by human vetters. We also provide a homogeneously classified set of TESS candidates suitable for future training.
rate research
Read More
A novel artificial intelligence (AI) technique that uses machine learning (ML) methodologies combines several algorithms, which were developed by ThetaRay, Inc., is applied to NASAs Transiting Exoplanets Survey Satellite (TESS) dataset to identify exoplanetary candidates. The AI/ML ThetaRay system is trained initially with Kepler exoplanetary data and validated with confirmed exoplanets before its application to TESS data. Existing and new features of the data, based on various observational parameters, are constructed and used in the AI/ML analysis by employing semi-supervised and unsupervised machine learning techniques. By the application of ThetaRay system to 10,803 light curves of threshold crossing events (TCEs) produced by the TESS mission, obtained from the Mikulski Archive for Space Telescopes, the algorithm yields about 50 targets for further analysis, and we uncover three new exoplanetary candidates by further manual vetting. This study demonstrates for the first time the successful application of the particular combined multiple AI/ML-based methodologies to a large astrophysical dataset for rapid automated classification of TCEs.
We present TRICERATOPS, a new Bayesian tool that can be used to vet and validate TESS Objects of Interest (TOIs). We test the tool on 68 TOIs that have been previously confirmed as planets or rejected as astrophysical false positives. By looking in the false positive probability (FPP) -- nearby false positive probability (NFPP) plane, we define criteria that TOIs must meet to be classified as validated planets (FPP < 0.015 and NFPP < 10^-3), likely planets (FPP < 0.5 and NFPP < 10^-3), and likely nearby false positives (NFPP > 10^-1). We apply this procedure on 384 unclassified TOIs and statistically validate 12, classify 125 as likely planets, and classify 52 as likely nearby false positives. Of the 12 statistically validated planets, 9 are newly validated. TRICERATOPS is currently the only TESS vetting and validation tool that models transits from nearby contaminant stars in addition to the target star. We therefore encourage use of this tool to prioritize follow-up observations that confirm bona fide planets and identify false positives originating from nearby stars.
We have adapted the algorithmic tools developed during the Kepler mission to vet the quality of transit-like signals for use on the K2 mission data. Using the four sets of publicly-available lightcurves on MAST, we produced a uniformly-vetted catalog of 772 transiting planet candidates from K2 as listed at the NASA Exoplanet archive in the K2 Table of Candidates. Our analysis marks 676 of these as planet candidates and 96 as false positives. All confirmed planets pass our vetting tests. 60 of our false positives are new identifications -- effectively doubling the overall number of astrophysical signals mimicking planetary transits in K2 data. Most of the targets listed as false positives in our catalog either show prominent secondary eclipses, transit depths suggesting a stellar companion instead of a planet, or significant photocenter shifts during transit. We packaged our tools into the open-source, automated vetting pipeline DAVE (Discovery and Vetting of Exoplanets) designed to streamline follow-up efforts by reducing the time and resources wasted observing targets that are likely false positives. DAVE will also be a valuable tool for analyzing planet candidates from NASAs TESS mission, where several guest-investigator programs will provide independent lightcurve sets -- and likely many more from the community. We are currently testing DAVE on recently-released TESS planet candidates and will present our results in a follow-up paper.
State of the art exoplanet transit surveys are producing ever increasing quantities of data. To make the best use of this resource, in detecting interesting planetary systems or in determining accurate planetary population statistics, requires new automated methods. Here we describe a machine learning algorithm that forms an integral part of the pipeline for the NGTS transit survey, demonstrating the efficacy of machine learning in selecting planetary candidates from multi-night ground based survey data. Our method uses a combination of random forests and self-organising-maps to rank planetary candidates, achieving an AUC score of 97.6% in ranking 12368 injected planets against 27496 false positives in the NGTS data. We build on past examples by using injected transit signals to form a training set, a necessary development for applying similar methods to upcoming surveys. We also make the texttt{autovet} code used to implement the algorithm publicly accessible. texttt{autovet} is designed to perform machine learned vetting of planetary candidates, and can utilise a variety of methods. The apparent robustness of machine learning techniques, whether on space-based or the qualitatively different ground-based data, highlights their importance to future surveys such as TESS and PLATO and the need to better understand their advantages and pitfalls in an exoplanetary context.
We present results of a study on identifying circumbinary planet candidates that produce multiple transits during one conjunction with eclipsing binary systems. The occurrence of these transits enables us to estimate the candidates orbital periods, which is crucial as the periods of the currently known transiting circumbinary planets are significantly longer than the typical observational baseline of TESS. Combined with the derived radii, it also provides valuable information needed for follow-up observations and subsequent confirmation of a large number of circumbinary planet candidates from TESS. Motivated by the discovery of the 1108-day circumbinary planet Kepler-1647, we show the application of this technique to four of Keplers circumbinary planets that produce such transits. Our results indicate that in systems where the circumbinary planet is on a low-eccentricity orbit, the estimated planetary orbital period is within <10-20% of the true value. This estimate is derived from photometric observations spanning less than 5% of the planets period, demonstrating the strong capability of the technique. Capitalizing on the current and future eclipsing binaries monitored by NASAs TESS mission, we estimate that hundreds of circumbinary planets candidates producing multiple transits during one conjunction will be detected in the TESS data. Such a large sample will enable statistical understanding of the population of planets orbiting binary stars and shed new light on their formation and evolution.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
