Learning sufficient low-dimensional structures through conditional optimal transport
arXiv:2607.18861
2026
Architecture
1 ideas extracted · analyzed Aug 30, 2026
What the math gives to ML
The paper provides a principled way to learn a low-dimensional covariate bottleneck that preserves the entire conditional response distribution, rather than only its mean or selected moments. Its transferable asset is the factorization of a conditional quadratic-optimal-transport map and its current-state velocity through a sufficient representation, turning sufficient dimension reduction into a conditional flow-matching objective. A neural implementation can jointly learn an encoder and a conditional velocity field using an empirical relaxed optimal-transport coupling. This is especially relevant for heteroscedastic, multimodal, or distribution-shifted targets where a mean-predicting bottleneck loses information.
Ideas from this paper
△ Mechanism confirmed, baseline not beaten
2026
Replace a conventional regression bottleneck with an encoder whose representation is trained to preserve the conditional law of the target through conditional optimal transport. The encoder produces a low-dimensional z, while a conditional velocity field transports a fixed reference distribution into the observed target distribution given z; minimizing flow-matching error forces z to retain multimodality, conditional variance, and other distributional information.
Useful7/10
Difficulty6/10
Novelty6/10