arXiv subjects

J. L. Fischer

Publications and source records attributed to J. L. Fischer.

3 recordsLinked to original sources

arXiv2019

Transfer learning for galaxy morphology from one survey to another

Deep Learning (DL) algorithms for morphological classification of galaxies have proven very successful, mimicking (or even improving) visual classifications. However, these algorithms rely on large training samples of labelled galaxies (typically thousands of them). A key question for using DL classifications in future Big Data surveys is how much of the knowledge acquired from an existing survey can be exported to a new dataset, i.e. if the features learned by the machines are meaningful for different data. We test the performance of DL models, trained with Sloan Digital Sky Survey (SDSS) data, on Dark Energy survey (DES) using images for a sample of $\sim$5000 galaxies with a similar redshift distribution to SDSS. Applying the models directly to DES data provides a reasonable global accuracy ($\sim$ 90%), but small completeness and purity values. A fast domain adaptation step, consisting in a further training with a small DES sample of galaxies ($\sim$500-300), is enough for obtaining an accuracy > 95% and a significant improvement in the completeness and purity values. This demonstrates that, once trained with a particular dataset, machines can quickly adapt to new instrument characteristics (e.g., PSF, seeing, depth), reducing by almost one order of magnitude the necessary training sample for morphological classification. Redshift evolution effects or significant depth differences are not taken into account in this study.

astro-ph.GA↗

arXiv2018

Improving galaxy morphologies for SDSS with Deep Learning

We present a morphological catalogue for $\sim$ 670,000 galaxies in the Sloan Digital Sky Survey in two flavours: T-Type, related to the Hubble sequence, and Galaxy Zoo 2 (GZ2 hereafter) classification scheme. By combining accurate existing visual classification catalogues with machine learning, we provide the largest and most accurate morphological catalogue up to date. The classifications are obtained with Deep Learning algorithms using Convolutional Neural Networks (CNNs). We use two visual classification catalogues, GZ2 and Nair & Abraham (2010), for training CNNs with colour images in order to obtain T-Types and a series of GZ2 type questions (disk/features, edge-on galaxies, bar signature, bulge prominence, roundness and mergers). We also provide an additional probability enabling a separation between pure elliptical (E) from S0, where the T-Type model is not so efficient. For the T-Type, our results show smaller offset and scatter than previous models trained with support vector machines. For the GZ2 type questions, our models have large accuracy (> 97\%), precision and recall values (> 90\%) when applied to a test sample with the same characteristics as the one used for training. The catalogue is publicly released with the paper.

astro-ph.GA↗

arXiv2014

A deficit of ultraluminous X-ray sources in luminous infrared galaxies

We present results from a Chandra study of ultraluminous X-ray sources (ULXs) in a sample of 17 nearby (D_L<60 Mpc) luminous infrared galaxies (LIRGs), selected to have star formation rates (SFRs) in excess of 7 M_sun yr^-1 and low foreground Galactic column densities (N_H < 5*10^20 cm^-2). A total of 53 ULXs were detected and we confirm that this is a complete catalogue of ULXs for the LIRG sample. We examine the evolution of ULX spectra with luminosity by stacking the spectra of individual objects in three luminosity bins, finding a distinct change in spectral index at luminosity ~2 *10^39 erg s^-1. This may be a change in spectrum as 10 M_sun black holes transit from a ~Eddington to a super-Eddington accretion regime, and is supported by a plausible detection of partially-ionised absorption imprinted on the spectrum of the luminous ULX (L_X ~5*10^39 erg s^-1) CXOU J024238.9-000055 in NGC 1068, consistent with the highly ionised massive wind that we would expect to see driven by a super-Eddington accretion flow. This sample shows a large deficit in the number of ULXs detected per unit SFR (0.2 ULXs M_sun^-1 yr^-1). This deficit also manifests itself as a lower differential X-ray luminosity function normalisation for the LIRG sample than for samples of other star forming galaxies. We show that it is unlikely that this deficit is a purely observational effect. Part of this deficit might be attributable to the high metallicity of the LIRGs impeding the production efficiency of ULXs and/or a lag between the star formation starting and the production of ULXs; however, we argue that the evidence -- including very low N_ULX/L_FIR, and an even lower ULX incidence in the central regions of the LIRGs -- shows that the main culprit for this deficit is likely to be the high column of gas and dust in these galaxies, that fuels the high SFR but also acts to obscure many ULXs from our view.

astro-ph.HE↗