patternsthat

Patternsthat is a term used to describe a class of patterns that encode their own generative rules. In this sense, a patternsthat consists of both the observable motif and the underlying rule set that produced it, allowing the pattern to be analyzed, compressed, or reproduced. The concept spans data-driven contexts, such as pattern mining and motif discovery, as well as design-oriented contexts, including textiles, graphics, and architecture. The term is not widely formalized in mainstream literature but appears in interdisciplinary discussions of generative patterning and computational creativity.

Key characteristics include hierarchical structure, self-similarity, modular composition, and rule-based generation. Patternsthat can be described with grammars, fractal rules, or procedural parameters, enabling scalable reproduction and manipulation. Detection methods combine grammar induction, pattern mining, and neural representations to infer both motifs and rules from observations.

Applications include efficient data compression by capturing generative rules, visual design workflows that explore parameterized motifs, procedural content generation in games, and anomaly detection by identifying deviations from expected generative rules. In practice, patternsthat support exploration of design spaces while preserving coherent structure, making it possible to interpolate between motifs or scale patterns without losing their character.

Examples include a tiling pattern generated by a simple substitution rule across scales; a leaf vein pattern produced by a branching grammar; a textile print formed from a small motif expanded through affine transformations.

See also: pattern mining, motif discovery, grammar induction, procedural generation, fractals.