background.tex

\section{Background}
\label{section:background}

In this section, we provide background knowledge about vulnerability and neural network which is also known as deep learning.

\subsection{Neural Network.}
Neural network is a model used in machine learning. Nodes in neural network are called neurons.
Connections between neurons are used to propagate values to each other, with value of each edge works as a weight of propagation.
Adjusting weights of each connection by propagating forward and backward, the neural network `learns' how to evaluate desired output features from features contained in input data.
Neural network can be structured as multiple layers and operates propagation between adjacent layers.
Model with more layers can represent more complex attributes, which are called deep learning models.

\subsection{Convolutional Neural Network.}
Convolutional neural network (CNN) is a deep learning model which is widely used in image and video recognition and natural language processing.
Typically, the CNN takes 2-dimensional data, usually image, as input and feed forward them in network with convolution operations.
A convolution operation takes a small region in input and reduce the region into single value.
The convolution layer works as a feature extractor, reducing the complexity of input by preserving desired high level features only.