Search of strong lens systems in the Dark Energy Survey using convolutional neural networks

We present our search for strong lens, galaxy-scale systems in the first data release of Dark Energy Survey (DES), based on a color-selected parent sample 18 745 029 luminous red galaxies (LRGs). used convolutional neural network (CNN) to grade this LRG with values between 0 (non-lens) and 1 (lens). Our training set mock lenses is data-driven, that is, it uses lensed sources taken from HST-COSMOS images lensing DES sample. A total 76 582 cutouts were obtained score above 0.9, which then visually inspected classified into two catalogs. The one contains 405 lens candidates, 90 clear features counterparts, while other 315 require more evidence, such as higher resolution imaging or spectra, be conclusive. 186 candidates are newly identified by search, 16 among most promising (best) candidates. second catalog includes 539 ring galaxy This will useful false positive future CNNs. For best we carry out color-based deblending source light without fitting any analytical profile data. method shown very efficient deblending, even compact objects complex morphology. Finally, selected 52 single deflector test an automated modeling pipeline has capacity successfully model 79% within acceptable computing runtime.