symmetry-discovery-questionnaire

Symmetry discovery guides ML teams through a structured domain-analysis workflow to uncover data invariances and equivariances, reducing data requirements and improving model generalization across various data modalities.

Uncover data symmetries by running a structured questionnaire that applies domain-specific checklists and transformation testing templates, generating a machine-readable summary of identified invariances and equivariances.

Yes, transformation testing templates support symmetry discovery across point clouds and graphs by applying domain-specific checklists to identify equivariance and invariance, helping guide domain analysis and physical-constraint identification.

Document physical constraints by encoding symmetry discovery steps, collected evidence, and an output rubric into a machine-readable format, supporting audit, validation, and reproducible symmetry discovery across modalities like physics simulations.

Symmetry discovery works for time series and tabular data by applying a structured domain-aware questionnaire and transformation tests to identify invariances, helping design models that respect these symmetries and improve generalization.