Class of Neural networks desgned to operate on data that can be represented as graph $G$ , where $G = (V , E)$ or $(Nodes , Edges)$ , yet graph data is so complex that its created alot of challenges for existing machine learning algorithms
Q : Why is it hard to analyze a graph ?
Convential Machine learning and Deep learning are specialized in simple data type like images with same structure and size. Text and Speech are sequences, so we can think of them as line graphs , But there are more complex graph without a fixed form.
In machine learning algorithms have a core assumptions that instances are independent of each other. this is false for graph data, because each node is related to others by links of various type
Graph Neural Networks (GNNs) are a class of neural networks designed to operate on data that can be represented as graphs. This includes social networks, citation networks, molecular structures, and many other types of data where relationships between entities are crucial. The working mechanism of GNNs can be broken down into several key components