Real-Life Industrial Process Data Mining -- Activities of
RISC-PROS

Paavo Nieminen

In my talk, I will introduce the on-going research project
RISC-PROS (Realtime Industrial Process and Sensor Network
Management Using Data Mining and Dimension Reduction). I hope to
make attendant colleagues better aware of this activity, the
theory and methods being developed, and possibilities of finding
mutual research interests. The project is a continuation,
hybridization, and application of earlier research done at the
University of Jyväskylä and at the Tel Aviv University. The main
interest is in three methods called 'Statistically robust
K-estimates clustering', 'Statistically robust training of
artificial neural networks with regularization', and 'Diffusion
Geometry Framework'. I will talk about the project's goals,
describe the real-life industrial environment, introduce the
computational methods, and discuss our current research interests
and questions in the project. There may be an incidental opinion
about how software should be shared and jointly developed while we
work with our different projects and studies.