This paper provides a comprehensive overview of Non-Intrusive Load Monitoring (NILM) systems and their associated methods for disaggregated energy sensing. The authors review the state-of-the-art load signatures and disaggregation algorithms used for appliance recognition, highlighting challenges and future research directions. The paper is structured into several sections, covering the general framework of NILM, appliance features for energy disaggregation, learning and inference in NILM systems, and performance evaluation of load disaggregation algorithms. Key topics include the differences between Intrusive Load Monitoring (ILM) and NILM, the importance of accurate energy monitoring for energy management, and the limitations of ILM such as high costs and installation complexity. The paper also discusses various methods for feature extraction, load identification, and system training, including supervised and unsupervised learning techniques. Finally, it provides a detailed comparison of different learning algorithms and their performance metrics, emphasizing the need for standardized evaluation methods to improve the accuracy and reliability of NILM systems.This paper provides a comprehensive overview of Non-Intrusive Load Monitoring (NILM) systems and their associated methods for disaggregated energy sensing. The authors review the state-of-the-art load signatures and disaggregation algorithms used for appliance recognition, highlighting challenges and future research directions. The paper is structured into several sections, covering the general framework of NILM, appliance features for energy disaggregation, learning and inference in NILM systems, and performance evaluation of load disaggregation algorithms. Key topics include the differences between Intrusive Load Monitoring (ILM) and NILM, the importance of accurate energy monitoring for energy management, and the limitations of ILM such as high costs and installation complexity. The paper also discusses various methods for feature extraction, load identification, and system training, including supervised and unsupervised learning techniques. Finally, it provides a detailed comparison of different learning algorithms and their performance metrics, emphasizing the need for standardized evaluation methods to improve the accuracy and reliability of NILM systems.