Abstract
Prediction and visualization have caused a great deal of interest in the research field, primarily as new graph datasets are becoming very popular. Graph convolutional networks are an area in Machine Learning in which graph datasets are trained and operated using neural networks. This work includes the application of the graph convolution networks model is implemented on a graph-structured dataset to describe the similarity between them. The GCN is implemented on a set of data that is represented in the graph datasets of protein types of different cancer cells. The graph datasets are the various patterns of proteins of various cancer-type cells that are set and represented in the form of nodes, labels, attributes, and indicators.
This work includes the implementation of graph convolutional networks GCN which helps to identify and compactly achieve objectives, on the protein data through hidden layers and neurons to check for the properties of the data set and determine their properties and matching assets. A neural network is applied to capture the dependencies, after which the graph convolutional layer is applied to operate and capture the dependencies present between the nodes. PyTorch, which is an open-source ML-based library is used for the implementation of this visualization technique. As an add-on, GCN has been implemented on a Cora dataset to classify its nodes into seven different layers.