Usage of Basic Geometrical Ideas and Principles of Symmetry for Machine Learning

The paper discusses the usage of geometric principles and theories in order to manipulate the position of decision boundary itself to classify data points belonging to two different classes. Moving away from the close relation between statistical approaches at classification, we will attempt to create a classifier that establishes a more direct connection, and works with the decision boundary. We explore binary classification problems using the dataset, compare convergence conditions and evaluate our results against some popular classifiers.