Problem Definition: This Master’s Thesis investigates how new sensors can be used in a recognition chain by training them with the existing ones. Thereby, the focus shall be the question how sufficient the dynamic transfer learning approach from a sensor jacket (or alternatively from one or more smart phones) to more unobtrusive sensors (e.g., wrist worn watch, sensor on the shoe), is. As no groundtruth is available, the thesis shall also compare and evaluate different methods to estimate the expected accuracy of the system, when training and using the new sensors. An existing framework is available, which shall be used within this Master’s thesis.
- Related Work Review
- Development and Validation of suitable signal processing and machine learning algorithms and autonomous realization of an experiment
- Evaluation and Experiment
- Interest in component-based software development (Java, OSGi)
- Good command of English (understanding of related work; Thesis is preferred written in English)
- Typesetting using LaTeX (→ Miktex, TeXnicCenter) + BibTeX (→ Jabref)
- Start: flexible, as of now
- Estimated completion date: within 6 months
- [KHF+2011a] Marc Kurz, Gerold Hölzl, Alois Ferscha, Alberto Calatroni, Daniel Roggen, Gerhard Tröster: Real-Time Transfer and Evaluation of Activity Recognition in an Opportunistic System, Third International Conference on Adaptive and Self-Adaptive Systems and Applications (ADAPTIVE2011), September 25-30, Rome, Italy, ISBN: 978-1-61208-156-4, pp. 73-78, 2011.
- [KHF+2011b] Marc Kurz, Gerold Hölzl, Alois Ferscha, Hesam Sagha, Jose Millán, Ricardo Chavarriaga: Dynamic Quantification of Activity Recognition Capabilities in Opportunistic Systems, Fourth Conference on Context Awareness for Proactive Systems: CAPS2011, 15-16 May 2011, Budapest, Hungary, 2011.
- [Calatroni2011] Alberto Calatroni, Daniel Roggen and Gerhard Tröster: Automatic Transfer of Activity Recognition Capabilities between Body-Worn Motion Sensors: Training Newcomers to Recognize Locomotion, Eighth International Conference on Networked Sensing Systems (INSS'11), 2011