I am planning to apply different machine learning methods to broad range of tasks, such as object classification based on both flux measurements and raw images, novelty detection and correction of the distorted images.
Broad spectroscopic lines, large redshift range and variety of properties make quasar detection in photometric surveys a particularly difficult task, and estimation of their photometric redshifts is even more challenging. I will present a quasar detection method based on photometric ugri data in Kilo-Degree Survey (KIDS) - an imaging deep and wide field survey covering 447 sq. deg. on the sky...
In a talk I would like to present methods developed or currently under development for TOROS project that have potential to be useful for LSST. Most important methods developed be me for TOROS is background rejection using Convolutional Neural Networks (with 99.5 % accuracy). The other methods currently under development involves galaxy subtraction using GANs, scheduling using reinforcement...
