Fundamentals of Whole‐System, Systemic, and Multiperspective Machine Learning
Parag A. Kulkarni · 2012
This chapter discusses the need for systemic learning and selectively using the learned information to produce the required results. Systemic learning includes the analysis of different dependencies and interdependencies and determining these leverage points dynamically. Another important aspect of learning is working on these leverage points. This chapter introduces the concept of selective systemic and whole systemic learning and further implementation of it in the real-life scenarios. The purpose of the system learning is to extend the time view and system view boundaries with learning and inference. In case of systemic machine learning, it is necessary to understand and define the system space for the decision problems. The data used for learning needs to represent whole-system features. Multiperspective learning is required for multiperspective decision making. Bayesian belief network or influence diagrams acknowledge the usefulness of the frameworks for addressing complex, dynamic real-life problems. Controlled Vocabulary Terms belief networks; learning (artificial intelligence)