The Maximum Euclidean Distance in a Class Defines the Boundary of Neighborhood and Leads to a New Machine Learning Algorithm | Chapter 06 | Recent Advances in Mathematical Research and Computer Science Vol. 7

 When we are given a data set, we assign a class to each data point based on the values and/or characteristics of attributes. In machine learning, the k-Nearest Neighbor (kNN) method is a relatively basic and powerful tool for performing this. It is built on the concept of data points belonging to the same class being neighbours. In kNN, one evaluates the Euclidean distances of the test data or unknown data from all the data points of all the classes in the training data to discover the class to which it belongs. The class to which test data or unknown data is closest the most number of times, out of the k nearest distances, where k is any number higher or equal to 1, is the class assigned to the test data or unidentified data. In this chapter, I offer an alternative to kNN, which I refer to as the ANN technique (Alternative Nearest Neighbor). The defining aspect of ANN that distinguishes it from kNN is the concept of neighbour. The unknown data is neighboured to the class whose maximum Euclidean distance from its data points is smaller than or equal to the maximum Euclidean distance between all of the class's training data points in ANN. As a result, each unknown data will always receive a unique solution from ANN. The solution in kNN, on the other hand, may vary depending on the number of nearest neighbours k. As a result, the performance of kNN may vary as k is changed. This is not the case with ANN, whose performance is tailored to a given training dataset.

The fundamental goal for developing the ANN machine learning method was to improve on the traditional kNN method by making it independent of the parameter k and eliminating the necessity for the user to select k neighbours based on experience or other criteria.

The ANN produces a 100 percent accurate result for the training data [1] used in this article.


Author(S) Details

Pushpam Kumar Sinha
Department of Mechanical Engineering, Netaji Subhas Institute of Technology, Amhara, Bihta, Patna, India.

View Book:- https://stm.bookpi.org/RAMRCS-V7/article/view/5491