Temporal and Object Relations in Unsupervised Plan and Activity Recognition.
Richard G. Freedman, Hee‐Tae Jung, Shlomo Zilberstein · 2015
We consider ways to improve the performance of unsuper-vised plan and activity recognition techniques by consid-ering temporal and object relations in addition to postural data. Temporal relationships can help recognize activities with cyclic structure and are often implicit because plans have degrees of ordering actions. Relations with objects can help disambiguate observed activities that otherwise share a user’s posture and position. We develop and investigate graphical models that extend the popular latent Dirichlet al-location approach with temporal and object relations, exam-ine the relative performance and runtime trade-offs using a standard dataset, and consider the cost/benefit trade-offs these extensions offer in the context of human-robot and human-computer interaction. 1