Semi Supervised Learning Using Segmentation and Clustering for Classification of Medical Images

Abstract

Semi supervised learning is deeply based on embedded clustering it helps to learn the feature representation by using various iterations with labeled and unlabeled data points to compute the target distribution and to predict the data. During this iterative prediction process the learning algorithm uses labeled samples in order to keep the consistent of the model with tuned labeling vice versa it also helps for the improvement of feature representation and prediction. Hue angle information are used to identify the high and low activity region. The system being employed converts the input images into grayscale and segments it for further classification.