Abstract:
Deep convolutional neural networks are compound functions of a certain form (i.e. based on a certain set of simple functions) that are dependent on a large number of tunable parameters and that are applicable to $n$-dimensional arrays, e.g. images. Over the last four years, deep convolutional neural networks have revolutionized computer vision, pattern recognition and machine learning. In the talk, I will discuss what are these neural networks, how they work, and how researchers are trying to visualize and understand what is learned inside the networks. Time permitting, I will also discuss how such networks can be used for image synthesis.
