Weighted Self-Organizing Maps: Incorporating User Feedback

One interesting way of accessing collections of multimedia objects is by methods of visualization and clustering. Growing self-organizing maps provide such a solution, which adapts automatically to the underlying database. Unfortunately, the result of the clustering greatly depends on the definition of the describing features and the used similarity measure. In this paper, we present a general approach to improve the obtained clustering by incorporating user feedback (in the form of drag-and-drop) into the underlying topology of the self-organizing map.