Rummage of Machine Learning Algorithms in Cancer Diagnosis

1. Introduction

Machine learning, a subfield of Artificial Intelligence, provides algorithms to learn from past experiments while performing a particular task and measuring performance. By working continuously on a task, the performance of the task gets improved and the user experience as well. A Machine learning system has a training data set working as knowledge base and rules for decision making (Blum 2007). Machine learning is the building and exploring of methods in a computer programming language and making them “learn”. The program developed using machine learning algorithms accesses the data, trains the machine and tests it again for performance evaluation. The most important characteristic of machine learning is its ability to forecast. In machine learning, a model for prediction is build by existing information and it is further used for predicting the data. The major aspect of learning is the features selection from the data set as all the features cannot be used in learning. The data set may have multiple fields and perspectives. The selection of features is done according to their relevance, implication and scenario (Du & Swamy 2013). The primary goal of machine learning is to produce and enhance the learning algorithms and models in order to facilitate their easy application in various disciplines such as agriculture (Patrício & Rieder 2018), banking (Erdogan 2013), cyber security (Buczak & Guven 2016), economics (Einav & Levin 2014), finance (Lin et al., 2012), insurance (Gan 2013), natural language processing (Collobert & Weston 2008), online & traditional marketing (Tripathy et al., 2006), healthcare (Crown 2015), network & telecommunication (Richter & Khoshgoftaar 2018) and others as well.