I ma enthusiast of all aspects of big data processing and all technologies that bring this enthusiasm from dream into reality – like CUDA. For over ten years I have been involved in many challenging projects related to image and laser data processing. I am one of the authors of automated traffic signs detection and recognition systems and bird-eye image mosaic creation process
When we look at the world our brain instantly turns the images we see into information, intuition and feeling. It does so with an enormous computational effort, using a network with overwhelming, unexplored architecture. We are not even close to create a machine of comparable cababilities, but step by step, improving both the hardware and algorithmic approach, we can make machine understand images they see.
In this talk you will learn how to design and develop such a network. We will go from the basic machine learning concepts, through the advanced practical tips, up to using the most-recent state-of-the art architectures in practice. You will see examples on how insanely complicated problems can be solved using a portable computer. And in the meanwhile you will understand what is going on, why new layers are added and why new concept have to be introduced.
While this presentation briefly introduces all the basic concepts, as we move on it quickly touches advanced and expert areas of the field.