A Novel Approach of Clustering Using Diverse Transfer Learning

Clustering is a way to discover groupings in a dataset without using supervised learning. It is an NP-hard problem to cluster data. In most cases, machine learning approaches are necessary to solve the information clustering issue. Affinity dissemination is one of the more recent ideas for cluster detection. The example principle underpins the affinity transmission framework. Similarities between data points are used as input in the affinity dissemination method. Using this approach, you can get a collection of data points that are good representations of the clusters. In order to create clusters, all non-exemplar information points are assigned to neighbouring exemplars, which are known as exemplars. In this paper, we proposed K-means based on Diverse Transfer Learning (DTL). The beginning of centres determines the clusters in the K-Means result. This problem is attempted to be solved using K-Means based on DTL. The DTL-based K-means algorithm begins by using K-Means to locate the text cluster centre. The following stage involves using these text centroids to establish the clustering using K-mean centres for images. Despite being just as sophisticated as K-Means, DTL-based K-Means produces better outcomes. We conducted experiments with the K-Means clustering approach and K-Means clustering approach based on Diverse Transfer Learning and determined their F-Measure, Clarity, and Entropy scores for comparison.