Keynotes

Abstract: In many „Internet of Things“ applications, sensor data plays a vital role for monitoring real-world phenomena. However, sensor data can have multiple data quality issues; to make things worse, these issues are context-dependent, i.e., they change over time. This keynote highlights some data management challenges that need to be addressed for future IoT infrastructures, with a focus on data quality in sensor-based applications. It illustrates these challenges with applications and research topics from the joint research programme „FutureIOT“, funded by the Bavarian Research Foundation and introduces the „Living Lab Bamberg“, an open platform for sensor-based research and long-running experiments.

Video: Keynote 1

Abstract: In this talk we will present the selfBACK decision support system (DSS) project that uses Case-Based Reasoning for supporting patients with low back pain. The DSS has been developed during the last 2 years and will now be tested with patients in a pilot study and a multisite randomized controlled trial. The selfBACK DSS will be used by the patient him/herself to facilitate, improve and reinforce self-management of LBP. Specifically, selfBACK is designed to assist the patient in deciding and reinforcing the appropriate actions to manage own low back pain after consulting a health care professional. The decision support is conveyed to the patient via a smartphone app in the form of advice for self-management. The advice is tailored to each patient based on the symptom state, symptom progression, the patients goal-setting, and a range of patient characteristics including information from a physical activity-detecting wristband worn by the patient.

Video: Keynote 2

Abstract: Machine-Learning ermöglicht die Entwicklung innovativer und neuartiger digitaler Produkte. Die Einbindung unterschiedlicher Datenquellen, deren intelligente Verknüpfung in Kombination mit selbstlernenden Mechanismen bilden hierfür die Grundlage. Doch absolut erfolgskritisch ist die Akzeptanz aus Anwendersicht. Aus der Akzeptanzforschung ist bekannt, dass neue und „freiwillige“ Software einen 10-fach höheren Nutzen als bestehende Lösungen aufweisen muss, um in den täglichen Gebrauch überzugehen. Frank Giesler stellt einen in der Praxis bewährten Ansatz vor, der durch enge Einbeziehung der späteren Nutzer eine erfolgreiche datengetriebene digitale Produkt­entwicklung sicherstellt.

Video: Keynote 3

Abstract: I will give an overview of some exciting recent developments in deep probabilistic modeling, which combines deep neural networks with probabilistic models for unsupervised learning. Deep probabilistic models are capable of synthesizing artificial data that highly resemble the real training data, and are able fool both machine learning classifiers as well as humans. These models have numerous applications in creative tasks, such as voice, image, or video synthesis and manipulation. At the same time, combining neural networks with strong priors results in flexible yet highly interpretable models for finding hidden structure in large data sets. I will summarize my group’s activities in this space, including measuring semantic shifts of individual words over hundreds of years, summarizing audience reactions to movies, and generating artificial video sequences with partial control over content and dynamics.

Video: Keynote 4