Activity recognition has become a key issue in smart home environments. The problem involves learning high level activities from low level sensor data. Activity recognition can depend on several variables; one such variable is duration of engagement with sensorised items or duration of intervals between sensor activations that can provide useful information about personal behaviour. In this paper a probabilistic learning algorithm is proposed that incorporates episode, time and duration information to determine inhabitant identity and the activity being undertaken from low level sensor data. Our results verify that incorporating duration information consistently improves the accuracy.
|Title of host publication||Unknown Host Publication|
|Number of pages||16|
|Publication status||Published - 1 Sep 2010|
|Event||KSEM 2010 - Belfast, UK|
Duration: 1 Sep 2010 → …
|Period||1/09/10 → …|