Workshop on Learning in Parallel Universes

Learning in Parallel Universes is a relatively new data mining research field; it refers to the setup of having different descriptor spaces in parallel, whereby each typically reflects different properties of the underlying objects. The learning task itself is two-fold: to exploit these multiple object representations and link structures that are shared among multiple universes (but not necessarily all – which is the objective of Multi-View Learning) and, secondly, to also identify structures that are specific to individual universes (due to the inherent nature of the respective universes). One classical example is the analysis of molecular data, whereby the molecules are described in various ways, for instance based on measures of their 3D layout, structural information or properties of their surface charge distribution. This learning problem can also be found in the domain of image analysis (where descriptor generation has been its own field of research ever since, resulting in descriptors based on color distribution, silhouettes, general meta information, etc.). Other examples include the mining of web pages, 3D objects or the categorization of pieces of music.

The workshop focus encompasses theories and application in this field. This includes novel approaches and extensions of existing solutions that specifically address the problem of learning in Parallel Universes.

Key Dates

  • Paper Submission due June 16th, 2008
  • Notifications of Acceptance on July 16th, 2008
  • Camera Ready Papers due August 14th, 2008
  • Workshop: September 19th, 2008

Organization

Program Committee

To be announced

Program

To be announced

Paper Submission

Authors are invited to submit manuscripts dealing with the topic of the workshop to lpu08-submission@inf.uni-konstanz.de. We welcome papers that present new results, ongoing projects or completed work (even previously published). We particularly invite submissions discussing specific application domains of Learning in Parallel Universes. Final versions of accepted papers will appear in the workshop proceedings. Submission implies the willingness of at least one of the authors to register and present the paper. The instructions for authors and the LaTeX packages can be found at http://www.springer.de/comp/lncs/authors.html. There is no limitation on paper length although we recommend the usual page limits of around 10 pages.