Multi-Sensor Data and Knowledge Fusion -- A Proposal for a Terminology Definition

Abstract

Fusion is a common tool for the analysis and utilization of available datasets and so an essential part of data mining and machine learning processes. However, a clear definition of the type of fusion is not always provided due to inconsistent literature. In the following, the process of fusion is defined depending on the fusion components and the abstraction level on which the fusion occurs. The focus in the first part of the paper at hand is on the clear definition of the terminology and the development of an appropriate ontology of the fusion components and the fusion level. In the second part, common fusion techniques are presented.

Fully probabilistic design for knowledge fusion between Bayesian filters under uniform disturbances

“…where E f denotes expectation with respect to f . FPD consists in minimizing this KLD (12) objective as a function of f (y t |x t , f S ) in (11), for the fixed ideal (10)…”

“…Recall that our aim is to specialize FPD-optimal Bayesian transfer learning (FPD-BTL framework) defined in (10), (11), (13) to a pair of Bayesian filters under bounded observational and state noises. We now address this aim.…”

“…Assume that the (unoptimized) variational target observation model, f (y t |x t , f S ) (11), is also uniform with bounded support. If X ∩ t = ∅, then the optimal choice of f (y…”

“…). After substitution into (11), the FPD-optimal joint pdf f o ∝ f (y t |x t ) f S (x t |d S (t − 1)), i.e. it is zero ∀x t ∈ X ∩ t .…”

“…Examples of successful deployment of transfer learning in data fusion are found in [7] - [10]. In accordance with the DIKW classification scheme proposed in [11], we will refer to transfer learningbased fusion as knowledge fusion.…”