Predicting Resident Intention Using Machine Learning

Abstract

The smart home environment is embedded with technologies like machine learning, deep learning artificial intelligence and internet of things. The services intended by the resident are provided by the smart home environment through interactions with the home appliances. In recent years, predicting user intention and behaviour in real-time applications like e-commerce, smart home, and entertainment, healthcare has appeared as a popular research domain. Contextual modalities such as speech, activity, emotion, object affordances, and physiological parameters are the features through which the intention of the resident can be predicted concerning the home appliances like door, television, light, etc. Other examples of contextual modalities include gesture and mood. Home appliances can be made smart by embedding intelligent algorithms which in turn helps them to understand the residents' intentions. This creates a dynamic relationship between the resident & home appliances thereby improving the resident’s satisfaction level. For example, the context: Resident issuing the command ‘OPEN’ by standing in front of the door illustrates that the intention is to get the door opened automatically. Contextual modalities are the main source for a system that works based on resident intention prediction. In this paper, an effort has been made to predict the resident intentions by applying machine learning algorithms like Decision Tree-based ID3, Naive Bayes classifier, and Rule-based classifier on context-aware door dataset.