Basics of Neural Network

This chapter systematically introduces the foundational models of neural networks, beginning with the basic building blocks of neurons and perceptrons. It then explains the role of activation functions before detailing three progressively complex network architectures tailored for specific data types. The text first covers the standard feedforward neural network (FNN), then introduces the convolutional neural network (CNN) as a specialized architecture that uses convolution and pooling to effectively process spatial data like images. Finally, it addresses the challenge of sequential information by presenting the recurrent neural network (RNN) and its advanced, memory-equipped variants, Long Short-Term Memory (LSTM) and the Gated Recurrent Unit (GRU), which are designed to capture temporal relationships in data.