Machine Learning-Assisted Device Selection in a Context-Sensitive Ubiquitous Multimodal Multimedia Computing System

In a computing system where a user moves from one environment to another, and as such the user's context and computing resources also change, a constant user intervention to enable/disable various devices to suit his needs is time-consuming and diminishes user productivity. Instead, a machine could be trained to acquire knowledge so that it would do the work (i.e. calculation and decision making) itself and leaves human do something else that is more important. In a ubiquitous multimodal multimedia (MM) computing system, the selection of appropriate media and modalities (i.e. devices) is based on user's context, user profile, and user's environment (a.k.a. pre-condition scenario). There are numerous possibilities of a pre-condition scenario and the available devices also changes depending on user's computing environment. Indeed, a machine learning (ML) component could be trained to "remember" all pre-condition scenarios, and each one's device selection (a.k.a. post-condition scenario). This ML component could also be trained to find a replacement to every missing or defective selected device. The ML component is integrated into a ubiquitous system making it available anytime, anywhere. This work is an original contribution in ML, one that permits automatic system adaptation based on user's environment