Machine Learning: Fundamentals

• Published in: Prognostics and Health Management of Electronics pp. 85 - 109
• Main Authors: Kang, Myeongsu, Jameson, Noel Jordan
• Format: Book Chapter
• Language: English
• Published: Chichester, UK Wiley 2018; John Wiley and Sons Ltd
• Edition: 1
• Series: Wiley - IEEE
• Subjects: batch learning; machine learning; online learning; probability theory; prognostics and health management; reinforcement learning; semi‐supervised learning; supervised learning; unsupervised learning; Training; Machine learning algorithms; Supervised learning; Machine learning; Task analysis; Prognostics and health management; Anomaly detection
• ISBN: 1119515335; 9781119515333
• DOI: 10.1002/9781119515326.ch4

Prognostics and health management (PHM) facilitates maintenance decision‐making and provides usage feedback for the product design and validation process. Electronic component and product manufacturers need new ways to gain insights from the massive volume of data recently streaming in from their systems and sensors, and this can be accomplished by using machine learning (ML). This chapter provides the fundamentals of ML. ML algorithms can be divided into the following four categories depending on the amount and type of supervision they need while training: supervised, unsupervised, semi‐supervised, and reinforcement learning. ML algorithms can be classified into two different learning methods based on whether or not the algorithms can learn incrementally from a stream of incoming data: batch and online learning. Probability theory plays a significant role in ML, specifically as the design of learning algorithms often depends on probabilistic assumption of the data.