Semi Supervised Learning Using Graph Data Structure – A Review
M A Aromal, Akhtar Rasool · 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV) · 2021
Graph based Semi-supervised learning (GSSL) is machine learning technique which can out-perform supervised classification when large number of unlabelled data to be classified with a small number of labelled data. In GSSL the given samples are taken to each vertices of a graph and the bridge connecting these vertices called edges having edge weight representing the similarity between the two vertices. Graph representation of data helps to convey the information among the data more efficiently. The basic idea of graph based semi supervised learning is transferring the label information from one sample to another in relationship with the similarity between them. In this paper some popular GSSL algorithms are discussed. Brief summary of these methods are also given. This review will provide researchers a better understanding in graph based semi-supervised learning.