Handwritten digit recognition of MNIST data using consensus clustering

Abstract

In Big Data, Pattern Recognition and Consensus Clustering techniques have growing importance to the academic and professional communities. Today there is a great concern for categorizing the data, as data in inappropriate category means inaccurate information, which in turn results wastage of resources and harming the organisation. Pattern recognition (PR) helps in avoiding poor categorization of data by identifying the correct structure of data in dataset. Recognizing a pattern is the automated process of finding the exact match and regularities of data, which is closely related to Artificial Intelligence and Machine Learning. PR acts as a primary step to provide clustering since it analyses the structure and vector value of each characters in dataset. Consensus Clustering (CC) also called as clustering ensembles, plays a significant role in categorizing and maintaining any type of data. This is a technique that combines multiple clustering solutions to obtain stable, accurate and novel results. In this paper, to implement PR and CC techniques, we use MNIST dataset which is a large database of handwritten digits that is commonly used for training various systems in the field of Machine Learning.