Vouw: Geometric Pattern Mining using the MDL Principle

TL;DR

This paper introduces Vouw, a heuristic algorithm for geometric pattern mining in matrices, leveraging the MDL principle to identify complex spatial structures with high-quality results demonstrated on synthetic benchmarks.

Contribution

The paper formalizes geometric pattern mining and proposes Vouw, a novel heuristic algorithm that effectively discovers complex spatial patterns using the MDL principle.

Findings

Vouw achieves high-quality pattern discovery on synthetic benchmarks.

The approach effectively captures complex spatial relations in geometric matrices.

Vouw demonstrates the potential of MDL-based pattern selection.

Abstract

We introduce geometric pattern mining, the problem of finding recurring local structure in discrete, geometric matrices. It differs from existing pattern mining problems by identifying complex spatial relations between elements, resulting in arbitrarily shaped patterns. After we formalise this new type of pattern mining, we propose an approach to selecting a set of patterns using the Minimum Description Length principle. We demonstrate the potential of our approach by introducing Vouw, a heuristic algorithm for mining exact geometric patterns. We show that Vouw delivers high-quality results with a synthetic benchmark.