Vouw: Geometric Pattern Mining using the MDL Principle

11/21/2019

∙

by   Micky Faas, et al.

∙

0

∙

share

We introduce geometric pattern mining, the problem of finding recurring local structure in raster-based data. It differs from other pattern mining problems by encoding complex spatial relations between elements, resulting in arbitrarily shaped patterns. After we formalise this new type of pattern mining, we discuss an approach to selecting a set of patterns using the Minimum Description Length principle. We demonstrate the viability of our approach by introducing Vouw, a heuristic algorithm that finds good solutions to a specific class of geometric pattern mining problems. We empirically show that Vouw delivers high-quality results by using a synthetic benchmark.