Mathematics and Machine Learning

Abstract

Machine learning is a branch of computer science that gives computers the ability to make predictions without explicitly being programmed. Machine learning enables computers to learn, as they process more and more data and make even more accurate predictions. Machine learning is becoming all pervasive in our daily lives, from speech recognition, medical diagnosis, customized content delivery, and product recommendations to advertisement placements to name a few. Knowingly or unknowingly, there is a very high chance that one would have encountered some form of machine learning several times in one’s daily activities. In cloud data centers, machine learning presents an opportunity to make systems autonomous and thus transforming data centers into those that are less error prone, secure, self tuning, and highly available. Mathematics forms the bedrock of machine learning. This paper aims at highlighting the concepts in mathematics that are essential for building machine learning systems. Topics in mathematics like linear algebra, probability theory and statistics, multivariate calculus, partial derivatives, and algorithmic optimizations are quintessential to implementing efficient machine learning systems. This paper will delve into a few of the aforementioned areas to bring out core concepts necessary for machine learning. Topics like principal component analysis, matrix computation, gradient descent algorithms are a few of them covered in this paper. This paper attempts to give the reader a panoramic view of the mathematical landscape of machine learning.