Color Space Transformation Network

Deep networks have become very popular over the past few years. The main reason for this widespread use is their excellent ability to learn and predict knowledge in a very easy and efficient way. Convolutional neural networks and auto-­‐encoders have become the normal in the area of imaging and computer vision achieving unprecedented BLOCKIN BLOCKIN accuracy BLOCKIN BLOCKIN levels BLOCKIN BLOCKIN in BLOCKIN BLOCKIN many BLOCKIN BLOCKIN applications. BLOCKIN BLOCKIN The BLOCKIN BLOCKIN most BLOCKIN BLOCKIN common BLOCKIN BLOCKIN strategy BLOCKIN BLOCKIN is to build and train networks with many layers by tuning their hyper-­‐parameters. While this approach has proven to be a successful way to build robust deep learning schemes BLOCKIN BLOCKIN it BLOCKIN BLOCKIN suffers BLOCKIN BLOCKIN from BLOCKIN BLOCKIN high BLOCKIN BLOCKIN complexity. BLOCKIN BLOCKIN In BLOCKIN BLOCKIN this BLOCKIN BLOCKIN paper BLOCKIN BLOCKIN we BLOCKIN BLOCKIN introduce BLOCKIN BLOCKIN a BLOCKIN BLOCKIN module BLOCKIN BLOCKIN that learns color space transformations within a network. Given a large dataset of colored images the color space transformation module tries to learn color space transformations that increase overall classification accuracy. This module has shown to increase overall accuracy for the same network design and to achieve faster BLOCKIN