Activity Recognition Based on Latent Knowledge Mining in Smart Home

Abstract

Activity recognition in smart home is one pattern recognition problem. Many activity recognition algorithms have appeared so far to recognize activities in smart home. Past researches have proved that dynamic and deep knowledge mining algorithms will help improve the accuracy. But because of the uncertainty of sensors and the complexity of the user activities, existing activity recognition methods still have a lot of room for improvement. Considering there is some latent knowledge existed in sensors or user activities, this paper proposes to recognize activities by exploring latent knowledge. Firstly, this paper improves activity recognition by extracting latent knowledge between sensors and activities, thereby proposed one feature preprocessing method. Then, it proves one new multi-resident activity recognition method based on latent knowledge in multi-resident activities. Simulations conclude that extracting latent knowledge can greatly enhance activity recognition.