Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
The paper introduces an entropy-adaptive Gumbel-Sinkhorn relaxation for learning permutations from structure without supervision. By locally modulating the Gumbel-Sinkhorn temperature based on assignment uncertainty, the method allows confident assignments to discretize early while preserving exploration for ambiguous ones. Experiments on sorting, jigsaw reconstruction, and routing tasks demonstrate improved training stability and permutation quality, especially with increasing problem size and ambiguity, compared to fixed-temperature baselines.
Forget global temperature tuning – this entropy-adaptive Gumbel-Sinkhorn lets you learn permutations from structure with greater stability and higher quality, even when things get ambiguous.
Many learning problems require uncovering a hidden ordering that reveals structure in unordered data, such as monotonicity in sorting or spatial continuity in jigsaw reconstruction. In these settings, permutations can be learned as latent operators by optimizing objectives defined directly on the reordered output, often without access to ground-truth orderings. Differentiable relaxations such as Gumbel-Sinkhorn make this approach practical by approximating permutation matrices with doubly stochastic matrices. However, learning from structure without supervision induces a non-uniform uncertainty: some assignments become confident early, while others remain ambiguous. Existing methods control this process using a single global temperature, forcing all assignments to sharpen or diffuse simultaneously and leading to instability at scale. We introduce an entropy-adaptive formulation of Gumbel-Sinkhorn that locally modulates temperature based on assignment uncertainty. This allows confident assignments to discretize early while preserving exploration where uncertainty remains. Across sorting and jigsaw reconstruction tasks and in routing-style settings, adaptive entropy control improves training stability and final permutation quality relative to fixed-temperature baselines, particularly as problem size and assignment ambiguity increase.