No abstract

No abstract

In this paper, we propose a multi-task active learning (AL) framework for an efficient characterization of buildings using features from multi-sensor earth observation data. Conventional AL methods establish query functions based on a preliminary trained learning machine to guide the selection of additional prior knowledge (i.e., labeled samples) for model improvement with respect to a single target variable. In contrast to that, here, we follow three multi-task AL meta-protocols to select unlabeled samples from a learning set which can be considered relevant with respect to multiple target variables. In particular, multi-task AL methods based on multi-variable criterion, alternating selection, rank combination, as well as hybrid approaches, which internalize multiple principles from the different meta-protocols, are introduced. Thereby, the alternating selection strategies implement a so-called one-sided selection (i.e., single-task AL selection for a reference target variable with simultaneous labeling of the residual target variables) with a changing leading variable in an iterative selection process. The multi-variable criterion-based methods and rank combination approaches aim to select unlabeled samples based on combined single-task selection decisions. Experimental results are obtained from two application scenarios for the city of Cologne, Germany. Thereby, the target variables to be predicted comprise building material type, building occupancy, urban typology, building type, and roof type. Comparative model accuracy evaluations underline the capability of the introduced methods to provide superior solutions with respect to one-sided selection and random sampling strategies.

Abstract-This paper presents a survey on the usage, opportunities and pitfalls of semantic technologies in the Internet of Things. The survey was conducted in the context of a semantic enterprise integration platform. In total we surveyed sixty-one individuals from industry and academia on their views and current usage of IoT technologies in general, and semantic technologies in particular. Our semantic enterprise integration platform aims for interoperability at a service level, as well as at a protocol level. Therefore, also questions regarding the use of application layer protocols, network layer protocols and management protocols were integrated into the survey. The survey suggests that there is still a lot of heterogeneity in IoT technologies, but first indications of the use of standardized protocols exist. Semantic technologies are being recognized as of potential use, mainly in the management of things and services. Nonetheless, the participants still see many obstacles which hinder the widespread use of semantic technologies: Firstly, a lack of training as traditional embedded programmers are not well aware of semantic technologies. Secondly, a lack of standardization in ontologies, which would enable interoperability and thirdly, a lack of good tooling support.

Abstract. In future, the so called "sensing enterprise", as part of the Future Internet, will play a crucial role in the success or the failure of an enterprise. We present our vision of an enterprise interacting with the physical world based on a retail scenario. One of the main challenges is the interoperability not only between the enterprise IT systems themselves, but also between these systems and the sensing devices. We will argue that semantically enriched service descriptions, the so called linked services will ease interoperability between two or more enterprises IT systems, and between enterprise systems and the physical environment.