This note is based on Coursera course by Andrew ng.
(It is just study note for me. It could be copied or awkward sometimes for sentence anything, because i am not native. But, i want to learn Deep Learning on English. So, everything will be bettter and better :))
The term, Deep Learning, refers to training Neural Networks, Sometimes very large Neural Networks. Let's start to the Housing Price Prediction example. I want to fit a function to predict the price of the houses, the function of size. The blue line is my function for predicting the price of the house as a function of this size. Then, the function that I've just fit the housing prices is a very simple neural network. If I draw neural network, it is like right to the graph.
Let's go more big steps. I have other features, which are the number of bedrooms, postal code and wealth. I may think one of things that the size and the number of bedrooms can affect the price of a house is family size. I may think postal code as walkability. And wealth may tell me how good is the school quality. So based on the size and number of bedrooms, i can estimate the family size, postal code, walkability, based on the postal code and wealth, i can estimate the school quality. Finally, i might think that the way people decide how much they are willing to pay for a house, is they look at the things.
So what i actually implement is this. I have a neural network with four inputs. And given these input features, the job of the neural network will be to predict y. And these circle in the middle are called hidden units in the neural network. Layer refers to input layer and hidden layer. The hidden layer is density connected because every input feature is connected to every one of these circles in the middle.
This image shows NN examples. The left is standard NN. Second one is convolutional NN, which is used for image data. Third one is recurrrent NN, which is very good for one-dimensional sequence data.
