Deep learning had gained lot of attention lately especially after AlphaGo beat Korean top GO play. Go is an ancient chess game invented in China and believe to have more possible combination than all molecules in the universal. Not only AlphaGo beat human player 4:1, there are also moves that are never observed before which suggest AI can innovate too.
In this workshop, we will go through some basic mathematics Neural Network which is the foundation of the Deep Learning. In those days in the early 2000s when I am doing my master thesis in "Wavelet Neural Network in Time Series Predication", we have to code every single computation and setup our own framework from ground up. Also at that time, the learning method computation, initialization method and etc are still in their infant. Currently, there are lot of frameworks such as Tensorflow, cafe and etc that make building of AI application much easier. The above mentioned learning method and initialization methods had also matured and become much more understood and stable.
However, I do not intend to go deep into the maths but just enough for us to have a good grab of the concept so that we wouldn't use the existing frameworks blindly.