Learning Unbiased Permutations via Flow Matching
About
Learning permutations is fundamental to sorting, ranking, and matching, but existing differentiable methods based on entropy-regularized Sinkhorn produce a single softened solution and collapse under ambiguity. We present PermFlow, a conditional flow matching framework that operates directly on the affine subspace of matrices with unit row and column sums. A closed-form tangent-space projector preserves these constraints exactly along every trajectory, by construction rather than through iterative correction, and a nearest-target coupling routes distinct noisy initializations toward distinct valid permutations. The result is a model that captures multimodal permutation distributions rather than collapsing them to a single mode. On a visual sorting task with blended-digit ambiguity and a symmetric linear assignment problem, PermFlow achieves high accuracy on unambiguous inputs and recovers both valid permutations under ambiguity, where Sinkhorn-based baselines structurally fail.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Sequence Sorting | NoisyMNIST (test) | Clean Accuracy98.1 | 11 | |
| Distribution Learning over Optimal Assignments | SLAP Bimodal N=20 2,000 instances (test) | Clean Accuracy94.6 | 8 | |
| Distribution Learning over Optimal Assignments | SLAP Bimodal (N=100) 2,000 instances (test) | Clean Accuracy82.8 | 8 |