Auto-RSM: An automated parameter-selection algorithm for the RSM map exoplanet detection algorithm

Dahlqvist, Carl-Henrik; Cantalloube, F.; Absil, Olivier

doi:10.1051/0004-6361/202141446

Cited by 5 publications

(15 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Considering the computational cost of the Auto-RSM framework, as well as the high degree of similarity observed between the optimal parametrisations of different ADI sequences (see Dahlqvist et al 2021b), we decided to rely on clustering to reduce the number of required optimisations. We divided our dataset into eight clusters using K-means clustering algorithm, based on parameters characterising the ADI sequence itself and the related observing conditions.…”

Section: Discussionmentioning

confidence: 99%

“…The underlying goals are to characterise the disks architecture and properties, and statistically link these properties to the stellar age, spectral type, and potential presence of companions. This paper contributes to the achievement of these objectives by applying the RSM detection algorithm (Dahlqvist et al 2021b) on the datasets, to detect potential planetary candidates. The RSM detection algorithm was designed to unveil point-like sources and is therefore not fitted to detect extended features such as debris disks.…”

Section: Survey Descriptionmentioning

confidence: 99%

“…In order to take full advantage of the RSM algorithm, we rely on the Auto-RSM optimisation framework (see Dahlqvist et al 2021b) to define the optimal sets of parameters for the PSFsubtraction techniques and the RSM algorithm itself. This optimisation pipeline being computationally expensive, we propose to apply it on a subset of targets representative of the whole dataset.…”

Section: Clusteringmentioning

confidence: 99%

“…We ran the Auto-RSM framework replacing the bottom-up approach by a top-down selection method to define the set of likelihoods cubes used to generate the final RSM detection maps. We also relied on the Auto-SNR framework (Dahlqvist et al 2021b) to generate optimised S/N maps. This framework uses the optimised parameters of the Auto-RSM framework for the PSF-subtraction techniques, but relies on a dedicated function to select and combine the optimal set of S/N maps.…”

Section: Point-source Characterisationmentioning

confidence: 99%

“…More recently, the Auto-RSM framework (Dahlqvist et al 2021b) was developed to reduce the burden of parameter selection and further optimise the performance of the RSM algorithm. This optimisation framework consists of three main steps: (i) the definition of the optimal set of parameters for the PSFsubtraction techniques, (ii) the optimisation of the RSM algorithm, and (iii) the selection of the optimal set of PSF-subtraction techniques and ADI sequences (Angular Differential Imaging, ) used to generate the final RSM probability map.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

The SHARDDS survey: limits on planet occurrence rates based on point sources analysis via the Auto-RSM framework

Dahlqvist

Milli

Absil

et al. 2022

A&A

Self Cite

View full text Add to dashboard Cite

Context. In the past decade, high contrast imaging allowed the detection and characterisation of exoplanets, brown dwarfs, and circumstellar disks. Large surveys provided new insights about the frequency and properties of massive sub-stellar companions with separations from 5 to 300 au. Aims. In this context, our study aims to detect and characterise potential exoplanets and brown dwarfs within debris disks, considering a diverse population of stars with respect to stellar age and spectral type. We present in this paper the analysis of a set of H-band images taken by the VLT/SPHERE instrument in the context of the SHARDDS survey. This survey gathers 55 main-sequence stars within 100 pc, known to host a high-infrared-excess debris disk, allowing us to potentially better understand the complex interactions between substellar companions and disks. Methods. We rely on the Auto-RSM framework to perform an in-depth analysis of the considered targets, via the computation of detection maps and contrast curves. A clustering approach is used to divide the set of targets into multiple subsets, in order to reduce the computation time by estimating a single optimal parametrisation for each considered subset. Detection maps generated with different approaches are used along with contrast curves to identify potential planetary companions. Planet detection and planet occurrence frequencies are derived from the generated contrast curves, relying on two well-known evolutionary models, namely AMES-DUSTY and AMES-COND. Finally, we study the influence of the observing conditions and observing sequence characteristics on the performance measured in terms of contrast.Results. The use of Auto-RSM allows us to reach high contrast at short separations, with a median contrast of 10 −5 at 300 mas, for a completeness level of 95%. A new planetary characterisation algorithm, based on the RSM framework, is developed and tested successfully, showing a higher astrometric and photometric precision for faint sources compared to standard approaches. Apart from the already known companion of HD206893 and two point-like sources around HD114082 which are most likely background stars, we did not detect any new companion around other stars. A correlation study between achievable contrasts and parameters characterising high contrast imaging sequences highlights the importance of the Strehl ratio, wind speed at a height of 30 meters, and presence of wind-driven halo to define the quality of high contrast images. Finally, planet detection and occurrence rate maps are generated and show, for the SHARDDS survey, a high sensitivity between 10 and 100 au for substellar companions with masses >10M J . Keywords. surveys-methods: data analysis-methods: statistical-techniques: image processing-techniques: high angular resolutionplanetary systems-planets and satellites: detection Based on observations collected at the European Southern Observatory under ESO programmes 096.C-0388(A) and 097.C-0394(A) F.R.S.-FNRS Senior Research Associate (e.g. Krivov 2010). These be...

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Survey Descriptionmentioning

confidence: 99%

Section: Clusteringmentioning

confidence: 99%

Section: Point-source Characterisationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

The SHARDDS survey: limits on planet occurrence rates based on point sources analysis via the Auto-RSM framework

Dahlqvist

Milli

Absil

et al. 2022

A&A

Self Cite

View full text Add to dashboard Cite

show abstract

NA-SODINN: A deep learning algorithm for exoplanet image detection based on residual noise regimes

Cantero,

Absil,

Dahlqvist

et al. 2023

A&A

View full text Add to dashboard Cite

Context. Supervised deep learning was recently introduced in high-contrast imaging (HCI) through the SODINN algorithm, a con-volutional neural network designed for exoplanet detection in angular differential imaging (ADI) datasets. The benchmarking of HCI algorithms within the Exoplanet Imaging Data Challenge (EIDC) showed that (i) SODINN can produce a high number of false positives in the final detection maps, and (ii) algorithms processing images in a more local manner perform better. Aims. This work aims to improve the SODINN detection performance by introducing new local processing approaches and adapting its learning process accordingly. Methods. We propose NA-SODINN, a new deep learning binary classifier based on a convolutional neural network (CNN) that better captures image noise correlations in ADI-processed frames by identifying noise regimes. The identification of these noise regimes is based on a novel technique, named PCA-pmaps, which allowed us to estimate the distance from the star in the image from which background noise started to dominate over residual speckle noise. NA-SODINN was also fed with local discriminators, such as signal-to-noise ratio (S/N) curves, which complement spatio-temporal feature maps during the model’s training. Results. Our new approach was tested against its predecessor, as well as two SODINN-based hybrid models and a more standard annular-PCA approach, through local receiving operating characteristics (ROC) analysis of ADI sequences from the VLT/SPHERE and Keck/NIRC-2 instruments. Results show that NA-SODINN enhances SODINN in both sensitivity and specificity, especially in the speckle-dominated noise regime. NA-SODINN is also benchmarked against the complete set of submitted detection algorithms in EIDC, in which we show that its final detection score matches or outperforms the most powerful detection algorithms. Conclusions. Throughout the supervised machine learning case, this study illustrates and reinforces the importance of adapting the task of detection to the local content of processed images.

show abstract

deep PACO: combining statistical models with deep learning for exoplanet detection and characterization in direct imaging at high contrast

Flasseur,

Bodrito,

Mairal

et al. 2023

Monthly Notices of the Royal Astronomical Society

View full text Add to dashboard Cite

Direct imaging is an active research topic in astronomy for the detection and the characterization of young substellar objects. The very high contrast between the host star and its companions makes the observations particularly challenging. In this context, post-processing methods combining several images recorded with the pupil tracking mode of telescope are needed. In previous works, we have presented a data-driven algorithm, PACO, capturing locally the spatial correlations of the data with a multivariate Gaussian model. PACO delivers better detection sensitivity and confidence than the standard post-processing methods of the field. However, there is room for improvement due to the approximate fidelity of the PACO statistical model to the time evolving observations. In this paper, we propose to combine the statistical model of PACO with supervised deep learning. The data are first pre-processed with the PACO framework to improve the stationarity and the contrast. A convolutional neural network (CNN) is then trained in a supervised fashion to detect the residual signature of synthetic sources. Finally, the trained network delivers a detection map. The photometry of detected sources is estimated by a second CNN. We apply the proposed approach to several data sets from the VLT/SPHERE instrument. Our results show that its detection stage performs significantly better than baseline methods (cADI and PCA), and leads to a contrast improvement up to half a magnitude compared to PACO. The characterization stage of the proposed method performs on average on par with or better than the comparative algorithms (PCA and PACO) for angular separation above 0.5 arcsec.

show abstract

Auto-RSM: An automated parameter-selection algorithm for the RSM map exoplanet detection algorithm

Cited by 5 publications

References 27 publications

The SHARDDS survey: limits on planet occurrence rates based on point sources analysis via the Auto-RSM framework

The SHARDDS survey: limits on planet occurrence rates based on point sources analysis via the Auto-RSM framework

NA-SODINN: A deep learning algorithm for exoplanet image detection based on residual noise regimes

deep PACO: combining statistical models with deep learning for exoplanet detection and characterization in direct imaging at high contrast

Contact Info

Product

Resources

About