Understanding Behaviors in Different Domains: The Role of Machine Learning Techniques and Network Science

Abstract

Recent developments in the Internet of Things (IoT), social media, and the data sciences have resulted in larger volumes of data than ever before, offering more opportunity for observing and understanding behaviors. Advances in data analytic and machine learning techniques have also enabled assessments to be more multi-faceted, incorporating data from more sources. Machine learning algorithms such as Decision Trees and Random Forests, K-nearest neighbors, and Artificial Neural Networks have been used to uncover hidden patterns in data and derive predictions and recommendations from a wide range of data types and sources. However, these do not necessarily yield insights into behaviors in complex systems/domains. Methods from mathematics such as Set Theory, Graph Theory, and Network Science may be useful in shedding light on the interactions and relationships within and across domains. This paper provides a description of the applications, strengths, and limitations of some of these techniques and methods.