Introduction to machine learning

Abstract

Machine learning is becoming increasingly popular in the neuroscientific literature. However, navigating the literature can easily become overwhelming, especially for the nonexpert. In this chapter, we provide an introduction to machine learning aimed at researchers, clinicians, and students with an interest in brain disorders, including psychiatry and neurology. We first provide a brief overview of how the most prominent theories of human learning from the fields of psychology and neuroscience influenced the development of modern cutting-edge machine learning methods. Second, we discuss how these methods differ from classical statistics and why they could be particularly suited to the investigation of brain disorders. In the final section of this chapter, we introduce a high-level taxonomy of the main approaches used in the machine learning literature: supervised learning, unsupervised learning, semisupervised learning, and reinforcement learning.