EPoS Contribution

Machine learning for filament detection Annie Zavagno LAM, Marseille, FR
Star formation occurs in filaments. Although extensively studied with both simulations and observations, their unbiased and systematic detection remains challenging. I will present a proof-of-concept study that shows how supervised deep learning can help in detecting filaments in 2D Galactic surveys. I will also present the perspective of this work, including the use of spectroscopic data cubes for a 3D detection of filament.

Caption: Example on an area of the Galactic Plane of the synthesis of the result obtained by the proof of concept published in A&A The top left image shows the area seen in near infrared emission (K band, 2MASS survey). This data was not used for training but is used here for the empirical validation of the result obtained through supervised learning and segmentation (bottom left image). This image shows the score map for a pixel to belong to the "filament" class, the structure we were trying to identify from the learning. The top right image shows the data used for this study, showing the column density distribution (amount of material on the line of sight) obtained from the Herschel space infrared satellite data. The black squares show the saturated areas where physical information cannot be obtained. The bottom right image shows the filaments known before our study, whose structures were used as masks for supervised learning using the convolutional networks Unet and Unet++.
Collaborators: L. Berthelot, LIS/LAM, FR T. Artières, LIS, FR P. Suin, LAM, FR D. Russeil, LAM, FR P. André, CEA, FR D. Arzoumanian, NAOJ, JP F.-X. Dupé, LIS, FR M. Gray, LAM, FR E. Schisano, IAPS-INAF, IT G. Li Causi, INAF, IT	Key publication Relevant topic(s): Feedback Filaments Triggered SF
Relevant Big Question: What is the impact of radiative feedback on star formation?