ABSTRACT
Recently, analysis of big data has gained significant traction in many fields including life sciences. Despite all statistical befits offered by this phenomenon, researchers demand new machine learning tools that can learn from this data in a reasonable period of time with limited computation power. In this chapter, we present a new mathematical framework, namely, active recursive Bayesian state estimation (Active-RBSE), that uses the Bayes rule to decouple measurements from different sources of input and utilizes active learning in a recursive state estimation paradigm to exploit learned statistics from earlier loops for including only informative instances of recorded (or to-be-recorded) signal. In this chapter, we have presented a particular objective function with a tractable solution for query optimization, but our framework bares the potential to employ different information theoretic objective functions based on requirements of the problem at hand. To illustrate the importance of information exploitation in query set design recursively, we employ a set of prerecorded EEG data, from an EEG-based brain–computer interface for typing, in a simulation study as a toy example, and the results suggest an improvement in terms of both typing accuracy and typing speed.
