"Interaction Modeling and Prediction in Smart Spaces: a Bio-Inspired Approach Based on Autobiographical Memory"

In Smart Spaces (SSs), the capability of learning from experience is fundamental for autonomous adaptation to environmental changes and for proactive interaction with users. New research trends for reaching this goal are based on neurophysiological observations of human brain structure and functioning. A learning technique that is used to provide the SS with the so-called Autobiographical Memory is presented here by drawing inspiration from a bio-inspired model of the interactions occurring between the system and the user. Starting from the hypothesis that user's actions have a direct influence on the internal system state variables and vice versa, a statistical voting algorithm is proposed for inferring the cause/effect relationships among users and the system. The main contribution of this paper lies in proposing a general framework that is able to allow the SS to be aware of its present state as well as of the behavior of its users and to be able to predict the expected consequences of user actions. In this paper, these concepts are explored in order to point out the relevant role of the structural coupling between the system and the environment (i.e., interaction with the user) in the development of context-aware SSs. In fact, a model is proposed here for learning in dynamic cognitive systems , which is able to understand the cause/effect relationships between the changes in system state and the environmental perturbations that are due to the presence of the user. The algorithm that is based on neurophysiological studies on how human self-consciousness arises and evolves is proposed to extract contextual information from heterogeneous sensor signals and to learn and predict interactions involving users that are present in an SS. The innovative aspect of this paper is the new way of modeling the interactions between the user and the system and its engineering implication in the development of context-aware learning/predicting strategies. To do so, multiple heterogeneous data coming from a set of sensors are jointly processed with the aim of detecting internal and external contextual events. Then, a nonparametric probabilistic interaction model is learned, which can be used to predict future events to be able to design anticipative decision strategies. The proposed mechanism, applied for processing and passively learning interactions between the system and the users, introduces new functionalities and modeling capabilities which can be exploited in SS design. This paper is organized as follows. In Section II, an approach for learning interactions between SSs and users, which is inspired by neurophysiological studies, is proposed. A procedure is introduced for using these memorized data to predict changes in the system’s internal status, which are caused by user interactions, potentially allowing self-reaction and self-adaptation capabilities. In Section III, the described algorithms are extensively tested in the scenario of an SS, which is installed in a university laboratory and which monitors the internal status of its devices and the external events produced by user actions. Section IV presents the comparisons with other learning approaches for AmI applications. Finally, in Section V, we conclude by commenting on open issues and possible future improvements.