Layered Markov models for complex temporal pattern recognition
Institute of Neural Information Processing
The University of Ulm
Time: 13:30-14:30 on Monday, November 19th, 2012
Place: Room 213, Engineering Blg.#8
New technical systems, so called Companion systems, aim at providing functionality in an individualized fashion by adapting to the user. To achieve this goal interdisciplinary research in psychology, neuro science and computer science is mandatory. The presentation highlights a methodological advancement in pattern recognition to enable the recognition of complex temporal user patterns. Based on a layered design of Markov models, the presented approach partitions the recognition task to patterns of different complexities and time granularities. The proposed architecture is introduced on the example of activity and affective user state recognition.
About the speaker
In the year of 2009 Michael Glodek achieved his Master degree in Neuro- and Bioinformatics at the university of Lübeck (Germany). He then moved to Ulm (Germany) and joined the German research collaboration “Companion-Technology for Cognitive Technical Systems” as a Ph.D. student. Within his work, he focuses on temporal and multi-view sensor fusion and the integration of sub-symbolic and symbolic information. As a result, his studies furthermore aim at detecting complex temporal patterns and incorporation of uncertain knowledge.
Prof. Toyoaki Nishida
Department of Intelligence Science and Technology
Graduate School of Informatics