Learning Automation Policies for Pervasive Computing Environments

摘要: If current trends in cellular phone technology, personal digital assistants, and wireless networking are indicative of the future, we can expect our environments to contain an abundance networked computational devices resources. We envision these acting orchestrated manner meet users' needs, pushing level interaction away from particular towards interactions with environment as a whole. Computation will be based not only on input explicitly provided by user, but also contextual information passively collected sensing devices. Configuring desired responses different situations need easy for users. However, anticipate that triggering many automation policies complex, unforeseen functions low-level information. This is problematic since users, though easily able perceive situations, define them devices' available information, even when such function (or close approximation) does exist. In this paper, present alternative approach specifying rules pervasive computing using machine learning techniques. Using approach, users generate training data policy through demonstration, and, after completed, learned employed future automation. enables automate changes combinations developed service within Gaia, system, deployed it prototype environment. were have demonstrate how sound lighting controls should adjust applications used environment, present, locations those then demonstrated preferences