Saved in:
Bibliographic Details
Main Authors: Lee, Dohoon, Park, Jaehyun, Kim, Hyunwoo J., Lee, Kyogu
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2404.14161
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911101733568512
author Lee, Dohoon
Park, Jaehyun
Kim, Hyunwoo J.
Lee, Kyogu
author_facet Lee, Dohoon
Park, Jaehyun
Kim, Hyunwoo J.
Lee, Kyogu
contents Flow and diffusion models have demonstrated strong performance and training stability across various tasks but lack two critical properties of simulation-based methods: freedom of dimensionality and adaptability to different inference trajectories. To address this limitation, we propose the Multidimensional Adaptive Coefficient (MAC), a plug-in module for flow and diffusion models that extends conventional unidimensional coefficients to multidimensional ones and enables inference trajectory-wise adaptation. MAC is trained via simulation-based feedback through adversarial refinement. Empirical results across diverse frameworks and datasets demonstrate that MAC enhances generative quality with high training efficiency. Consequently, our work offers a new perspective on inference trajectory optimality, encouraging future research to move beyond vector field design and to leverage training-efficient, simulation-based optimization.
format Preprint
id arxiv_https___arxiv_org_abs_2404_14161
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Multidimensional Adaptive Coefficient for Inference Trajectory Optimization in Flow and Diffusion
Lee, Dohoon
Park, Jaehyun
Kim, Hyunwoo J.
Lee, Kyogu
Machine Learning
Artificial Intelligence
I.2.6; I.5.1; F.1.1
Flow and diffusion models have demonstrated strong performance and training stability across various tasks but lack two critical properties of simulation-based methods: freedom of dimensionality and adaptability to different inference trajectories. To address this limitation, we propose the Multidimensional Adaptive Coefficient (MAC), a plug-in module for flow and diffusion models that extends conventional unidimensional coefficients to multidimensional ones and enables inference trajectory-wise adaptation. MAC is trained via simulation-based feedback through adversarial refinement. Empirical results across diverse frameworks and datasets demonstrate that MAC enhances generative quality with high training efficiency. Consequently, our work offers a new perspective on inference trajectory optimality, encouraging future research to move beyond vector field design and to leverage training-efficient, simulation-based optimization.
title Multidimensional Adaptive Coefficient for Inference Trajectory Optimization in Flow and Diffusion
topic Machine Learning
Artificial Intelligence
I.2.6; I.5.1; F.1.1
url https://arxiv.org/abs/2404.14161