Machine learning for data science: Mathematical or computational

Abstract

Machine learning usually requires getting a training and testing set of samples. The training set is used to obtain the model, and then, the testing set is used to verify the model. In general, a machine learning method requires an iterated process for reaching a goal. Machine learning is one of the research areas in artificial intelligence. Machine learning is mainly used to solve problems in classification and clustering. The distinction is that machine learning uses automated algorithms to learn from sample data for finding rules or making classifications. Therefore, the earliest machine learning method should be the regression method in statistics. Modern machine learning does not rely on a mathematical model such as linear equations used in regression. Online data mining and networking based applications nowadays request that machine learning be able to make decisions based on partial data sets. When more data samples are available, the algorithm must be able to adjust accordingly. Therefore, in cloud computing, and BigData related methods in data science, machine learning becomes the primary technology. We have introduced the PCA, k-NN and k-means, and other methods in artificial intelligence in Chap. 2. In this chapter, we will do an overview of other important machine learning methods such as decision trees, neural networks, and genetic algorithms. We will also introduce variational learning, support vector machine, and computational learning theory with some problems related to mathematical data processing.