Toy Benchmark Problem

The dataset used in the paper is a toy benchmark problem, where the data is generated by uniform sampling from a compact symmetry group G and other independent factors of variation V.

BibTex: