Relation Analysis for Semi-Lattice Based Multi-Activities Recognition Structure

A semi-lattice based activity model has been introduced to infer human daily activity within smart environment. For the purpose of improving the accuracy of activity recognition as well as to further study the layer function in lattice structure, we've explored the relationship between sensor and activity, and the relationship between different activities in this paper. Through the analysis of relations we can know that the minimal requirement of sensor/object for a lattice based recognizing model is three. Moreover we find out that the lattice structure for multi-activity recognition can minimize the overlapping between different activities and obtain a comparable accuracy with the process of single activity recognition.