Research map / Learning theory and algorithms / Optimal transport

Efficient One-to-many Domain Translation via Diffusive Entropic Optimal Transport

NeurIPS 2026 · Learning theory and algorithms › Optimal transport · approach: Transport map estimation

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes a diffusion-based approach to entropic optimal transport for one-to-many domain translation that avoids direct evaluation of generator log densities.

Problem

Domain translation must balance realism, input-output alignment, and output diversity; entropic optimal transport formalizes this trade-off, but maximizing entropy of a stochastic transport plan requires log-density evaluation of generator-induced distributions, which scales poorly with dimension.

Approach

Uses diffusion models to handle the entropy term of entropic optimal transport, sidestepping explicit log-density evaluation of the generator-induced distribution to make one-to-many translation efficient in high dimensions.

Experimental setup

Builds on: entropic optimal transport

Closest papers by meaning

• Discrete Diffusion Modeling by Estimating the Ratios of the Data Distribution (ICML 2024 · similarity 0.87)
• EVODiff: Entropy-aware Variance Optimized Diffusion Inference (NeurIPS 2025 · similarity 0.87)
• Diffusion Models are Secretly Exchangeable: Parallelizing DDPMs via Auto Speculation (ICML 2025 · similarity 0.87)
• Domain Guidance: A Simple Transfer Approach for a Pre-trained Diffusion Model (ICLR 2025 · similarity 0.87)
• Encoder-Decoder Diffusion Language Models for Efficient Training and Inference (NeurIPS 2025 · similarity 0.86)
• SemanticDLM+: Improving Diffusion LLMs through Bias-variance Trade-off in Transition Kernel Design (NeurIPS 2026 · similarity 0.86)