Challenges for Galaxy Morphological Classification from Large Surveys

Morphology is a key ingredient in the process of selecting a sample of galaxies for studying the physical mechanisms responsible for shaping the galaxies as we observe today. In the context of Big Data, the morphological classification of galaxies requires high performance for fast processing of images made available from large surveys (e.g. SDSS7, KiDS and upcoming instruments as LSST and Euclid.). A high performance in this scientific context requires expertise in the three computational segments: software, hardware and people ware. In this talk we will situate the galaxy morphology within the context of Big data (volume, velocity, variety, value and virtuosity), emphasizing the software performance that is currently available. We will also discuss which bottlenecks exist in the hardware segment and what are some strategies to optimize automatic classification considering new parallelism architectures and their adequacy to machine and deep learning techniques. We will show preliminar