 |
Anthony Cohn, University of Leeds , UK
Mining Video Data: Learning about Activities
In this talk I will present ongoing work at Leeds on building models of
video activity. I will present techniques, both supervised and
unsupervised, for learning the spatio-temporal structure of tasks and
events from video or other sensor data. In both cases, the
representation will exploit qualititive spatio-temporal relations. A
novel method for robustly transforming video data to qualitative
relations will be presented. For supervised learning I will show how the
supervisory burden can be reduced using what we term "deictic
supervision", whilst in the unsupervised case I will present a method
for learning the most likely interpretation of the training data. I will
also show how objects can be "functionally categorised" according to
their spatio-temporal behaviour and how the use of type information can
help in the learning process, especially in the presence of noise. I
will present results from several domains including a kitchen scenario
and an aircraft apron.
|
 |
Ian Horrocks, Oxford University, UK
Ontology Languages and Engineering
Ontologies and ontology based systems are rapidly becoming mainstream
technologies, with RDF and OWL now being deployed in diverse application
domains, and with major technology vendors starting to augment their
existing systems with ontological reasoning. For example, Oracle Inc.
recently enhanced its well-known database management system with modules
that use RDF/OWL ontologies to support "semantic data management", and
their product brochure lists numerous application areas that can benefit
from this technology, including Enterprise Information Integration,
Knowledge Mining, Finance, Compliance Management and Life Science
Research. The design of the high quality ontologies needed to support
such applications is, however, still extremely challenging. In this talk
I will describe the design of OWL, show how it facilitates the
development of ontology engineering tools, describe the increasingly
wide range of available tools, and explain how such tools can be used to
support the entire design, deployment and maintenance ontology
life-cycle.
|
 |
Thierry Denoeux, Université de Technologie de Compiègne, France
Theory of belief functions for data analysis and machine learning
applications: review and prospects
The Dempster-Shafer theory of belief functions provides a unified
framework for handling both aleatory uncertainty, arising from
statistical variability in populations, and epistemic uncertainty,
arising from incompleteness of knowledge. An overview of both the
fundamentals and some recent developments in this theory will first be
presented. Several applications in data analysis and machine learning
will then be reviewed, including learning under partial supervision,
multi-label classification, ensemble clustering and the treatment of
pairwise comparisons in sensory or preference analysis.
|
