Header And Logo
Site Navigation
Dr.-Ing. Boris Brandherm
|
||||
|
Address:
Saarland University
Building E 1 1
, Room 124
|
Projects |
Back to Top |
- READY-p1 1996-01-01 – 1998-12-31
- BAIR 2005-01-01 – 2007-12-31
- READY-p3 2002-01-01 – 2004-12-31
- READY-p2 1999-01-01 – 2001-12-31
Publications |
Back to Top |
2012
Sustainable Instrumentation of Everyday Commodities - Concepts and Tools
Boris Brandherm, Alexander Kröner, Jens Haupert, Michael Schmitz, Frank Lehmann and Ralf Gampfer
10th Annual IEEE International Conference on Pervasive Computing and Communications. IEEE International Conference on Pervasive Computing and Communications (PerCom-12), 10th, March 19-23, Lugano, Switzerland
[BibTeX]
Nachhaltiges Konzept zur Förderung der Therapietreue [Sustainable Concept for Increasing Compliance
Boris Brandherm, Alexander Kröner, Michael Schmitz, Jens Haupert, Frank Lehmann and Ralf Gampfer
Demographischer Wandel - Assistenzsysteme aus der Forschung in den Markt. Deutscher AAL-Kongress (AAL-12), 5. Deutscher AAL-Kongress mit Ausstellung, January 23-25, Berlin, Germany
[BibTeX]
2011
IUI 2011 Workshop on Location Awareness for Mixed and Dual Reality ( LAMDa )
Gerrit Kahl, Boris Brandherm, Tim Schwartz, Andreas Forsblom, Eyal Dim and Petteri Nurmi
2011 International Conference on Intelligent User Interfaces
Patientenindividuelle Förderung der Therapietreue durch intelligente Medikamentenverpackungen
Boris Brandherm, Alexander Kröner, Michael Schneider, Jens Haupert and Michael Schmitz
Demographischer Wandel - Assistenzsysteme aus der Forschung in den Markt. Deutscher AAL-Kongress (AAL), 4. Deutscher AAL-Kongress mit Ausstellung, January 25-26, Berlin, Germany
[BibTeX]
Synchronized Realities
Christoph Stahl, Jochen Frey, Jan Alexandersson, Boris Brandherm
Journal of Ambient Intelligence and Smart Environments, Vol. 3, Nr. 1/2011, pp. 13-25, IOS Press.
show abstract
Ambient Intelligence (AmI) promises future environments that support the well-being of their inhabitants through comfort and assistive environments. For the elderly users, Ambient Assisted Living (AAL) scenarios are targeting safe and independent living at home. Especially the integration of home automation components, such as intelligent light, window, and climate controls seem promising for such scenarios. In this paper, we present a new development method that is based on a detailed three-dimensional model and show how the model can be used for designing and evaluating assistive environments. Such a system has been developed and deployed. We present an architectural foundation of how Universal Remote Console technology can be applied to synchronize the virtual environment model with the real world. Furthermore, we extend previous work on Dual Reality by introducing the concept of Synchronized Realities that includes the synchronization between remote AmI environments. We will discuss application scenarios and show two demonstrators that we have implemented based on the proposed development method.
2010
Roles and Rights Management Concept With Identification by Electronic Identity Card
Boris Brandherm, Jens Haupert, Alexander Kröner, Michael Schmitz, Frank Lehmann
8th Annual IEEE International Conference on Pervasive Computing and Communications. IEEE International Conference on Pervasive Computing and Communications (PerCom-2010), March 29 - April 2, Mannheim, Germany, Pages 768-771, ISBN 978-1-4244-5328-3, IEEE Computer Society, 2010.
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
Adherence to therapy, what is termed "compliance", is a very important topic in Medicare. It has considerable effects not only in terms of enhancing quality of life for patients, but also in reducing the financial burden on the health care system. The individualized weekly medicament blister for patients (7x4 Box) from 7x4 Pharma increases the compliance and helps to keep the health expenditures as low as possible. A potential extension of the 7x4 Box could be a digital product memory which makes the imprinted data on the box electronically accessible in order to enable new innovative services which helps to increase the compliance. If we do not secure the data from an unauthorized access there is a possibility that third parties can read out the data unperceived and on a grand scale. In this paper we show how we can realize an authorized access -- individually or role-based -- on the data via identification by the new German electronic identity card.
Demo: Authorized Access on and Interaction With Digital Product Memories
Boris Brandherm, Jens Haupert, Alexander Kröner, Michael Schmitz, Frank Lehmann
8th Annual IEEE International Conference on Pervasive Computing and Communications. IEEE International Conference on Pervasive Computing and Communications (PerCom-2010), March 29 - April 2, Mannheim, Germany, Pages 838-840, ISBN 978-1-4244-5328-3, IEEE Computer Society, 2010.
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
In the Project "Semantic Product Memory" we exploit the lessons learned so far for the general design of a Digital Product Memory (DPM). DPMs can be used along their product's life cycle to collect data by different instances. In case of the individualized weekly medicament blister for patients from 7x4 Pharma not only the collected data but even the product itself is private. If we equip such a product with a DPM the collected data should be accessible -individually or role-based- only by authorized persons. This authorized access can be accomplished via a roles and rights management system and a secure identification provided by the new German electronic identity card. With our demonstrator we will present a system, which allows an authorized access on a DPM. Depending on the rights and roles of a user different data and different views will be available and displayed accordingly.
2009
A Simulation Framework for Sensor-Based Systems in Second Life
Helmut Prendinger, and Boris Brandherm, and Sebastian Ullrich
Mel Slater; Janet Weisenberger (Hrsg.). Presence: Teleoperators and Virtual Environments, Vol. 18, No. 6, Pages 468-477, The MIT Press, 12/2009.
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
This paper describes a simulation framework for sensor-based systems based on Second Life, a popular virtual 3D multi-user online world. With this platform, the components of sensor-based systems can be mapped to, or represented by, virtual devices and objects. The intuitive user interface of Second Life (SL) and its comprehensive visualization capability support evaluation tasks of ubiquitous computing applications. Developers, as avatars, can directly control and manipulate virtual devices. In this way, different settings of sensor-based systems can be tested, and interactively improved. A bidirectional interface between sensor-based systems and Second Life has been implemented to demonstrate Second Life as a testbed for an RFID-based positioning system.
Global Lab: an interaction, simulation, and experimentation platform based on Second Life and OpenSimulator
Anette von Kapri, Sebastian Ullrich, Boris Brandherm, Helmut Prendinger
Proceedings Pacific-Rim Symposium on Image and Video Technology (PSIVT'09), Tokyo, Japan
[ DOWNLOAD NOTE ]
show abstract
In this paper we describe a novel platform for virtual worlds called Global Lab. Through example applications, we demonstrate the capabilities of this platform for ubiquitous simulation and experimentation for sensor-based systems. Furthermore, with multimedia information representation and support for bi-directional communication real and virtual devices are connected in the example of a participatory ecosystem scenario. The Global Lab contains multimodal presentation and interaction techniques which enhance universal communication. The overall motivation is to provide a framework with strong support for collaboration in virtual worlds for research purposes.
2008
Simulation Framework in Second Life with Evaluation Functionality for Sensor-based Systems
Boris Brandherm, Sebastian Ullrich, and Helmut Prendinger
Proceedings of the 2nd Workshop on Ubiquitous Systems Evaluation (USE ’08). Seoul, South Korea, September 21, 2008. CEUR Workshop Proceedings, ISSN 1613-0073
[ DOWNLOAD NOTE ]
show abstract
This paper describes a simulation framework for sensor-based systems utilizing “Second Life”, a popular virtual three-dimensional multi-user online world. With this platform, the components of a sensor-based system can be mapped to (or, represented by) virtual counterparts. The intuitive user interface of Second Life and its comprehensive visualization support evaluation tasks of ubiquitous computing applications. Developers can directly control and manipulate virtual counterparts of real devices. In this way, different settings of a sensor-based system can be tested. The main contribution of our work consists of a bi-directional interface between sensor-based systems and Second Life.
Simulation Framework with Testbed for Sensor-based Systems in Second Life
Sebastian Ullrich, Boris Brandherm, and Helmut Prendinger
Demo Session of the 10th International Conference on Ubiquitous Computing (UBICOMP’08), 2008
show abstract
This paper introduces a simulation framework and testbed for sensor-based systems utilizing “Second Life”, a popular virtual three-dimensional multi-user online world. The main contribution of our work consists of a bi-directional interface to Second Life, which can greatly benefit developers of sensor-based systems for ubiquitous computing applications. While simulators for individual sensors exist, testing a complete system is still a major challenge. Our approach aims to solve this problem by proposing a framework that embeds complex sensor-based systems into Second Life, and utilize its versatile and intuitive three-dimensional user interface.
Dynamic Bayesian Network Based Interest Estimation for Visual Attentive Presentation Agents
Boris Brandherm, Helmut Prendinger, Mitsuru Ishizuka
Proceedings of 7th International Conference on Autonomous Agents and Multiagent Systems (AAMAS’08), ACM Press, 2008, 191-198
[ DOWNLOAD NOTE ]
show abstract
In this paper, we report on an interactive system and the results of a formal user study that was carried out with the aim of comparing two approaches to estimating users’ interest in a multi-modal presentation based on their eye gaze. The scenario consists of a virtual showroom where two 3D agents present product items in an entertaining way, and adapt their performance according to users’ (in)attentiveness. In order to infer users’ attention and visual interest with regard to interface objects, our system analyzes eye movements in real-time. Interest detection algorithms used in previous research determine an object of interest based on the time that eye gaze dwells on that object. However, this kind of algorithm does not seem to be well suited for dynamic presentations where the goal is to assess the user’s focus of attention with regard to a dynamically changing presentation. Here, the current context of the object of interest has to be considered, i. e., whether the visual object is part of (or contributes to) the current presentation content or not. Therefore, we propose to estimate the interest (or non-interest) of a user by means of dynamic Bayesian networks that may take into account the current context of the attention receiving object. In this way, the presentation agents can provide timely and appropriate response. The benefits of our approach will be demonstrated both theoretically and empirically.
Simulation of Sensor-based Tracking in Second Life (Demo Paper)
Boris Brandherm, Sebastian Ullrich, Helmut Prendinger
Proceedings of 7th International Conference on Autonomous Agents and Multiagent Systems (AAMAS’08), ACM Press, 2008, 1689-1690
[ DOWNLOAD NOTE ]
show abstract
This paper describes “Second Life” as a novel type of testbed and simulation environment for sensor-based applications. Second Life is a popular virtual online world that provides a free networked multi-user three-dimensional (3D) environment. The overall goal of our work is to support the development, testing, and deployment of sensor-based applications. In particular, pervasive systems like smart environments make heavy use of wireless sensor networks. However, the development of such systems requires much effort and the success of a system relies heavily on good planning and testing. Many different factors have to be taken into consideration and the environment has to be modeled carefully to foresee potential problems or to be able to perform changes before actual implementation. Until now, only custom-made solutions exist whereby technical limitations restrict adequate testing. By contrast, our approach introduces a flexible architecture for an extensible testbed for sensor-based applications. It employs Second Life to model an easily customizable three-dimensional environment with various interaction possibilities.
2007
Interest Estimation Based on Dynamic Bayesian Networks for Visual Attentive Presentation Agents
Boris Brandherm, Helmut Prendinger, Mitsuru Ishizuka
ICMI ’07: Proceedings of the 9th international conference on Multimodal interfaces, ACM, 2007, 346-349
[ DOWNLOAD NOTE ]
show abstract
In this paper, we describe an interface consisting of a virtual showroom where a team of two highly realistic 3D agents presents product items in an entertaining and attractive way. The presentation flow adapts to users’ attentiveness, or lack thereof, and may thus provide a more personalized and user-attractive experience of the presentation. In order to infer users’ attention and visual interest regarding interface objects, our system analyzes eye movements in real-time. Interest detection algorithms used in previous research determine an object of interest based on the time that eye gaze dwells on that object. However, this kind of algorithm is not well suited for dynamic presentations where the goal is to assess the user’s focus of attention regarding a dynamically changing presentation. Here, the current context of the object of attention has to be considered, i. e., whether the visual object is part of (or contributes to) the current presentation content or not. Therefore, we propose a new approach that estimates the interest (or non-interest) of a user by means of dynamic Bayesian networks. Each of a predefined set of visual objects has a dynamic Bayesian network assigned to it, which calculates the current interest of the user in this object. The estimation takes into account (1) each new gaze point, (2) the current context of the object, and (3) preceding estimations of the object itself. Based on these estimations the presentation agents can provide timely and appropriate response.
2006
Eingebettete dynamische Bayessche Netze n-ter Ordnung
Boris Brandherm
[ DOWNLOAD NOTE ]
show abstract
Das Ziel dieser Arbeit war die Konzeption, die Realisation und die Anwendung
eines Systems, das eingebettete Systeme mit geringer Rechenleistung und wenig
Arbeitsspeicher mit der Fähigkeit ausstattet, probabilistische Prozesse verarbeiten
zu können.
Die Grundlage für dieses System bildet der differentielle Ansatz von Darwiche zur Lösung Bayesscher Netze, der ein Bayessches Netz in ein multivariates Polynom umwandelt und dann auswertet. Diesen Ansatz von Darwiche haben wir so erweitert, dass nun auch die speziellen Bedürfnisse dynamischer Bayesscher Netze berücksichtigt werden.
Wir können im dynamischen Bayesschen Netz eine Vorwärts- und Rückwärtspropagierung und auch eine Kombination der beiden Verfahren durchführen. Alte Zeitscheiben und andere überflüssige Netzstrukturen können ohne Informationsverlust aufgerollt werden, wobei das Polynom bei konstantem Speicherverbrauch ausgewertet wird.
Aufbauend auf dieser theoretischen Ausarbeitung wurde eine Anwendung mit dem
Namen JavaDBN entwickelt, die dynamische Bayessche Netze in spezielle
Polynome umwandelt und für diese Quellcode (wahlweise Java oder C++) generiert.
Dieser Quellcode führt die Berechnungen für die Auswertung des Polynoms (die
Inferenz im dynamischen Bayesschen Netz) und das Anhängen neuer Zeitscheiben
mit dem gleichzeitigen Rollup (dem Abschneiden der vorhergehenden Zeitscheibe)
durch.
Der Quellcode wurde so realisiert, dass in der Initialisierungsphase alle
notwendigen Variablen eingeführt werden, und danach während der Laufzeit
kein Speicher mehr allokiert oder deallokiert werden muss. Somit wird auch
keine automatische Speicherbereinigung benötigt, die gerade in Echtzeit-Systemen
wie z.B. eingebetteten Systemen nicht toleriert werden kann, da der Zeitpunkt ihrer
Durchführung oft nicht vorhergesehen werden kann, und dadurch die Programmausführung zu nicht voraussehbaren Zeitpunkten unterbrochen werden würde.
Zusätzlich lässt sich dadurch der Speicherbedarf und die Laufzeit des Programmes nach oben abschätzen.
Ein weiterer großer Vorteil des Quellcodes liegt darin, dass er leicht lesbar ist,
so dass das Verfahren der Inferenz und des Rollups für den Anwender transparent bleibt. Dadurch lassen sich Veränderungen im Quellcode vornehmen, so dass sich Zusatzfunktionen wie beispielsweise Datenbankzugriffe oder Methodenaufrufe realisieren lassen, die nicht oder nur sehr schwer möglich wären, wenn der Inferenzalgorithmus vorkompiliert ist und das dynamische Bayessche Netz zur Verarbeitung als Eingabe erhält.
Dazu haben wir gezeigt, wie sich konstante Tabelleneinträge zu variablen Tabelleneinträgen umwandeln lassen, indem Arrays durch Methodenaufrufe ersetzt werden. Dies ist z.B. dann sinnvoll, wenn Tabelleneinträge nicht im Voraus bekannt sind oder sich zur Laufzeit verändern können. Ein Methodenaufruf könnte beispielsweise einen Datenbankzugriff realisieren (um beispielsweise die Tabelleneinträge aus einem Benutzerprofil einzulesen) oder von der
abgelaufenen Zeit seit der letzten Instantiierung einer Zeitscheiben abhängig sein.
Weiterhin erzeugt JavaDBN Quellcode für die Sensitivitätsanalyse und unterstützt das
Werkzeug MatLab.
Von den Anwendungen, die mit JavaDBN erzeugt wurden, möchte wir an dieser Stelle den Alarm Manager und den multisensoriellen Positionierungsservice Loriot erwähnen.
Anhand der prototypischen Anwendung Alarm Manager demonstrierten wir erstmals, wie physiologische Daten eines Benutzers durch physiologische Sensoren erfasst und durch ein dynamisches Bayessches Netz in Echtzeit auf einem persönlichen eingebetteten System interpretiert werden können, um dem Anwender Nachrichten (in diesem Fall eine wichtige Benachrichtigung) an seinen aktuellen Zustand angepasst zu präsentieren. Als Sensoren verwendeten wir einen Beschleunigungssensor, einen Muskeltätigkeitssensor und einen Augenbrauensensor.
In der Anwendung Loriot führen wir als Innovation die sogenannten
georeferenzierten dynamischen Bayesschen Netze (geoDBNs) ein, die es
ermöglichen, dass die Benutzerposition auf dem persönlichen Handheld-Gerät
berechnet werden kann, ohne dass eine externe Verbindung zu einem Server besteht,
auf dem dann die Positionsberechnung durchgeführt würde. Diese neue Methode hat
sowohl an Rechenzeit als auch Arbeitsspeicher nur einen geringen Ressourcenbedarf
und ist hoch-skalierbar für die In- und Outdoor-Positionierung. Momentan werden
IR- und RFID-Sensoren verwendet, aber ebenso lassen sich leicht andere Sensoren
einbauen, indem das dynamische Bayessche Netz um den entsprechenden Sensorknoten erweitert wird.
Für die Modellierung dynamischer Bayesscher Netze spezifizieren wir neue
Modellierungsstrukturen im Zusammenhang mit der Benutzermodellierung und
der Sensorverarbeitung und führen damit den Begriff der dynamischen Bayesschen
Netze n-ter Ordnung ein. Wir zeigen insbesondere im Zusammenhang mit der
Sensorverarbeitung wie sich die Probleme lösen lassen, die durch die Latenz
der Sensordaten auftreten können.
2005
Calculation of the User-Direction in an Always Best Positioned Mobile Localization System
Tim Schwartz, Boris Brandherm and Dominik Heckmann
Proceedings of the International Workshop on Artificial Intelligence in Mobile Systems (AIMS 2005), Salzburg, Austria, 2005
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
In an Always Best Positioned (ABP) localization system the output of different localization techniques are fused together to get an even better position accuracy. Besides the information about the position of a user, his viewing or walking direction is also important. This paper describes an extension of our mobile APB system that uses RFID tags and infrared beacons. We describe how different direction information – derived from different sensors or calculations – can be fused together with the help of Dynamic Bayesian networks.
Geo Referenced Dynamic Bayesian Networks for User Positioning on Mobile Systems
Boris Brandherm, Tim Schwartz
Proceedings of the International Workshop on Location- and Context-Awareness (LoCA 2005), Munich, Germany, 2005
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
The knowledge of the position of a user is valuable for a broad range of applications in the field of pervasive computing. Different techniques have been developed to cope with the problem of uncertainty, noisy sensors, and sensor fusion.
In this paper we present a method, which is efficient in time- and space-complexity, and that provides a high scalability for in- and outdoor-positioning. The so-called geo referenced dynamic Bayesian networks enable the calculation of a user’s position on his own small hand-held device (e.g., Pocket PC) without a connection to an external server. Thus, privacy issues are considered and completely in the hand of the user.
GUMO - the General User Model Ontology
Dominik Heckmann, Tim Schwartz, Boris Brandherm, Michael Schmitz and Margeritta von Wilamowitz-Moellendorff
Proceedings of the 10th International Conference on User Modeling (UM'2005), Edinburgh, UK, 2005, Springer-Verlag Berlin Heidelberg, LNAI 3538, pp. 428-432
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
We introduce the general user model ontology GUMO for the uniform interpretation of distributed user models in intelligent semantic web enriched environments. We discuss design decisions, show the relation to the user model markup language UserML and present the integration of ubiquitous applications with the user model service U2M.
Decentralized User Modeling with UserML and GUMO
Dominik Heckmann, Tim Schwartz, Boris Brandherm, Alexander Kröner
Proceedings of the Workshop on Decentralized, Agent Based and Social Approaches to User Modelling (DASUM 2005), Edinburgh, Scotland, 2005, pp. 61-65
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
We present a new architecture for decentralized user modeling and
briefly discuss the user model markup language USERML, the general user model
ontology GUMO for the uniform interpretation of decentralized user models, and
the integration of ubiquitous applications with the u2m.org user model service.
The motivation is that ubiquitous evaluation of user behavior with a variety of
systems in the web or the physical world might lead to attractive new services.
Recognition of Time Pressure via Physiological Sensors: Is the User's Motion a Help or a Hindrance?
Margeritta von Wilamowitz-Moellendorff, Christian Müller, Anthony Jameson, Boris Brandherm, Tim Schwartz
Proceedings of the Workshop "Adapting the Interaction Style to Affective Factors" (UM'2005), Edinburgh, UK, 2005
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
The recognition of a user's internal states via physiological sensors
is sometimes seen as a matter of detecting the direct physiological
correlates of the internal states. This type of detection can be
problematic when a user is moving around, as is often the case with
today's mobile systems. We present a study which illustrates that
detection of internal states is sometimes actually easier when the
subject is moving: The affective state may be associated with overt
behavior that results in detectable changes in the physiological variables.
Using Physiological Signals in a User-Adaptive Personal Assistant
Boris Brandherm, Holger Schultheis, Margeritta von Wilamowitz-Moellendorff, Tim Schwartz, Michael Schmitz
Proceedings of the 11th International Conference on Human-Computer Interaction (HCII-2005), Las Vegas, Nevada, USA, 2005
[BibTeX]
show abstract
Since psychophysiological signals are continuously available and usually quickly reflect changes of the user’s state, they constitute an important source of information for adaptive assistance systems. Despite their potential benefits however, physiological information is frequently neglected in current adaptive systems. This may—at least partly—be due to the fact that physiological measures cannot be easily used for adaptation. Instead several steps have to be taken to be able to draw on the advantages of physiology. First, each measure has to be evaluated regarding its suitability to distinguish between user states. Second, the exact relationships between physiological measures and states need to be identified. Finally, psychophysiological information has to be integrated over time and with other sources of information. In the scope of the BAIR project all three sub problems have been tackled and the respective solutions have been combined to give a systematic approach for the utilization of physiological information in user-adaptive personal assistance systems.
Navigational- and Shopping Assistance on the Basis of User Interactions in Intelligent Environments
C. Stahl, J. Baus, B. Brandherm, M. Schmitz and T. Schwartz
Proceedings of the IEE International Workshop on Intelligent Environments (IE 2005), University of Essex, Colchester, UK, 2005
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
This paper presents an overview about ongoing work in the project REAL, where we have set up the Saarland University Pervasive Instrumented Environment (SUPIE). In particular we introduce the intelligent environment’s architecture, which serves as the basis for different services and applications running in the environment and supporting their users in different tasks. On the basis of this information we outline our user and location-modeling component needed to establish the navigational– and shopping-assistants developed so far. Both assistants support their users with especially customized presentations. These presentations will be automatically scheduled and presented on public displays in the environment, as explained in the remarks about the presentation manager. Finally, we provide a short outlook on planned future work in the project.
User Adaptation on the Basis of Biosensors
[Benutzeradaption auf der Basis von Biosensoren]
Boris Brandherm
CeBIT 2005, Halle 9, Stand D09
[ DOWNLOAD NOTE ]
show abstract
As part of the collaborative research program 378 “Resource-Adaptive Cognitive Processes”, the project BAIR will show a method for user adaptation based on biosensors (BABB). To demonstrate the interpretability of biosensors, a system that allows one to control applications by bio signals is shown – in our case a computer game. The bio signals are gathered by sensors which are worn by the user, and include sensors for the acquisition of electrodermal activity, muscle tension, and eye movement. BABB uses dynamic Bayesian networks to filter out measurement errors, to combine sensor data (sensor fusion), and to estimate the result that also takes previous measurements into account. The thereby achieved complexity enables the system to deduce a variety of inferences – in our case the control of the computer game. As a vision, these techniques will be used in personal assistants, e.g. they will offer advice while shopping or traveling. As an example of what the near future will provide, these digital assistants will use bio sensors to determine time pressure or cognitive load of their users.
2004
Presentation of a Modular Framework for Interpretation of Sensor Data With Dynamic Bayesian Networks on Mobile Devices
Boris Brandherm and Michael Schmitz
12th GI-Workshop on Adaptivity and User Modeling in Interactive Software Systems (ABIS-2004), Berlin, 2004
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
We show in this work an example of how physiological data of a user acquired by bio sensors and interpreted by dynamic Bayesian networks on a personal mobile device are used to adapt notifications to the user state. The described sample application is an instance of our framework that is currently under development in our research group, aiming towards a modular and cross-platform toolkit for sensor integration.
We will give a brief overview of the whole system with its components and conclude with an outlook on future work in this project.
An Extension of the Differential Approach for Bayesian Network Inference to Dynamic Bayesian Networks
Boris Brandherm, Anthony Jameson
Special Issue "Uncertain Reasoning", International Journal of Intelligent Systems, Volume 19 Issue 8 (Part 1), 2004
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
We extend the differential approach to inference in Bayesian networks (BNs) (Darwiche, 2000) to handle specific problems that arise in the context of dynamic Bayesian networks (DBNs). We first summarize Darwiche's approach for BNs, which involves the representation of a BN in terms of a multivariate polynomial. We then show how procedures for the computation of corresponding polynomials for DBNs can be derived. These procedures permit not only an exact roll-up of old time slices but also a constant-space evaluation of DBNs. The method is applicable to both forward and backward propagation, and it does not presuppose that each time slice of the DBN has the same structure. It is compatible with approximative methods for roll-up and evaluation of DBNs. Finally, we discuss further ways of improving efficiency, referring as an example to a mobile system in which the computation is distributed over a normal workstation and a resource-limited mobile device.
2003
Adapting Spoken and Visual Output for a Pedestrian Navigation System, based on given Situational Statements
Rainer Wasinger, Dominika Oliver, Dominik Heckmann, Bettina Braun, Boris Brandherm, Christoph Stahl
11th GI-Workshop on Adaptivity and User Modeling in Interactive Software Systems (ABIS-2003), Karlsruhe, 2003, pp. 343-346.
[pdf]
[BibTeX]
show abstract
As mobile devices become more and more complex, there is an increasing desire for these devices to adapt to their users. This paper identifies parameters for different input
sources (user, device and environment), and the parameters of media output (speech, graphics, sound and text), that may be modified to tailor user presentation in a pedestrian navigation system. We also provide an initial insight into some of the causal relationships between our input and output parameters, with a specific focus on the effects that speech can contribute to the presentation of media output.
An Extension of the Differential Approach for Bayesian Network Inference to Dynamic Bayesian Networks
Boris Brandherm
Proceedings of the Sixteenth International FLAIRS Conference (FLAIRS-03), St. Augustine, Florida, USA, 2003
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
We extend the differential approach to inference in Bayesian networks (BNs) to handle specific problems that arise in the context of dynamic Bayesian networks (DBNs). We first summarize Darwiche's approach for BNs, which involves the representation of a BN in terms of a multivariate polynomial. We then show how procedures for the computation of corresponding polynomials for DBNs can be derived. These procedures permit not only an exact roll-up of old time slices but also a constant-space evaluation of DBNs. The method is applicable to both forward and backward propagation, and it does not presuppose that each time slice of the DBN has the same structure. It is compatible with approximative methods for roll-up and evaluation of DBNs.
2002
Efficient Modeling of Temporally Variable User Properties With Dynamic Bayesian Networks
Boris Brandherm
Proceedings of the Eighteenth National Conference on Artificial Intelligence (AAAI-02), Doctoral Consortium, Edmonton, Alberta, Canada, 2002
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
Efficiency issues are considered to handle dynamic Bayesian networks which estimate a user's resource limitations on the basis of symptomatic evidence in order to realize a situational presentation of information. Depending on the kind of data and on the frequency it arrives different methods of preprocessing this data must be applied to incorporate it as evidence in the DBN.
Empirically Grounded Decision-Theoretic Adaptation to Situation-Dependent Resource Limitations
Thorsten Bohnenberger, Boris Brandherm, Barbara Großmann-Hutter, Dominik Heckmann, Frank Wittig
Special Issue "Adaptivity and User Modeling", KI-Journal, Vol.3, 2002, pages 10-16
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
This article summarizes research on several interrelated general issues that can arise in the design and development of user modeling systems: the learning and subsequent adaptation of general user models on the basis of empirical data; the modeling of temporally variable properties of users, in particular time pressure and cognitive load; and the user-adaptive planning of interactions under uncertainty. The methods and results are integrated and illustrated with a prototype of a mobile assistance system for travelers in an airport.
2001
Verarbeitung von Sensordaten in Dynamischen Bayesschen Netzen
Boris Brandherm
9th GI-Workshop on Adaptivity and User Modeling in Interactive Software Systems (ABIS-2001), Dortmund, 2001
[ DOWNLOAD NOTE ]
[BibTeX]
show abstract
In diesem Beitrag werden Überlegungen vorgestellt, wie sich Daten aus Bio- und Umgebungssensoren zur Bewertung der Ressourcenlage einer Person (Zeitdruck und Arbeitsgedächtniskapazität) einsetzen lassen. Dazu wird einerseits ein Bayessches Netz für ausgewählte Biodaten modelliert und andererseits wird gezeigt, daß bei Kombination von physischen und verhaltensabhängigen Symptomen die Zeitscheiben in Dynamischen Bayesschen Netzen auf verschiedene Art und Weise instantiiert werden müssen.
2000
Rollup-Verfahren für komplexe dynamische Bayessche Netze
Boris Brandherm
Diplomarbeit, Fachbereich Informatik, Universität des Saarlandes, September 2000
[pdf]
[BibTeX]
show abstract
Vorwort
Bayessche Netze stellen einen Formalismus zur Verarbeitung und Repräsentation von unsicherem Wissen dar. Mit ihrer Hilfe lassen sich Domänen modellieren in denen mit Unsicherheit behaftete Schlußfolgerungen gezogen werden müssen.
Für das Lösen (Evaluieren) von Bayesschen Netzen wurden exakte und näherungsweise Verfahren (Inferenzalgorithmen) entwickelt. In beiden Fällen ist die Berechnungskomplexität von großer Bedeutung (Das exakte oder auch näherungsweise Lösen von Bayesschen Netzen ist NP-vollständig.).
Normalerweise beschreiben Bayessche Netze Kausalzusammenhänge, die entweder zeitunabhängig sind oder zu einem bestimmten Zeitpunkt gelten. Dynamische Bayessche Netze sind eine Erweiterung von Bayesschen Netzen und ermöglichen die Modellierung von zeitabhängigen Prozessen, indem über die Zeit sogenannte Zeitscheiben (Das sind Bayessche Netze, die durch Kanten mit dem alten Netz verbunden werden.) an das bestehende Netz angehängt werden können. Dabei wächst das Netz und für die Evaluation des Netzes werden die Laufzeiten immer länger und der Arbeitsspeicherbedarf immer umfangreicher. Damit die Evaluation eines Dynamischen Bayesschen Netzes in einem zeitlichen Rahmen bleibt, und auch der Arbeitsspeicherbedarf nicht über die vorhandenen Ressourcen steigt, müssen alte Zeitscheiben abgehängt werden, was auch als Rollup bezeichnet wird. [...]
Lectures |
Back to Top |
Linking Virtual and Real World through Android Applications
(SS 2011)
Linking Virtual and Real World through Android Applications
(SS 2011)