Title page of the tutorial on support vector machines by Alexey Nefedov

Support Vector Machines: A Simple Tutorial


This tutorial on support vector machines (SVM) provides a simple introduction to the method, easily accessible to anyone who has basic background in mathematics. It grew from a collection of notes and slides that I had been using since 2009 to present SVM to various groups in academia and industry. From repeated discussions with colleagues and students who were using SVM in various real-life applications, I got a feeling that despite a wealth of introductory texts about SVM, many people (especially those without strong background in mathematics) found it difficult to understand some important details of the method. My attempts to clarify those details eventually developed into this tutorial, where I tried to provide insights into the key aspects of the method, and give extended explanation of the math that lies in its foundation.
(PDF, 656 KB)

About the author

Alexey Nefedov, Ph.D., is a Senior Scientist at Merck Research Labs in Pennsylvania. He is using his expertise in bioinformatics, data analysis, machine learning and statistics to develop biomarkers that help to advance the discovery and development of novel drugs, therapies and vaccines. Prior to joining Merck in 2012, he worked in various academic and industrial institutions, including University of Pennsylvania, DIMACS, University of Liverpool, and Samsung Research Center. Over the last ten years, he used SVM in image recognition applications, veterinary epidemiology, mass spectrometry, and biomarker development.
Alexey’s LinkedIn profile

Excerpts from tutorial

Primary optimization problem for SVM, linearly separable classes
Primary optimization problem for SVM, linearly nonseparable classes
Parameter C

Selected resources and references on SVM

Contact form