Aligning Latent Spaces with Flow Priors

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Yizhuo, Ge, Yuying, Ge, Yixiao, Shan, Ying, Luo, Ping
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914394293665792
author Li, Yizhuo
Ge, Yuying
Ge, Yixiao
Shan, Ying
Luo, Ping
author_facet Li, Yizhuo
Ge, Yuying
Ge, Yixiao
Shan, Ying
Luo, Ping
contents This paper presents a novel framework for aligning learnable latent spaces to arbitrary target distributions by leveraging flow-based generative models as priors. Our method first pretrains a flow model on the target features to capture the underlying distribution. This fixed flow model subsequently regularizes the latent space via an alignment loss, which reformulates the flow matching objective to treat the latents as optimization targets. We formally prove that minimizing this alignment loss establishes a computationally tractable surrogate objective for maximizing a variational lower bound on the log-likelihood of latents under the target distribution. Notably, the proposed method eliminates computationally expensive likelihood evaluations and avoids ODE solving during optimization. As a proof of concept, we demonstrate in a controlled setting that the alignment loss landscape closely approximates the negative log-likelihood of the target distribution. We further validate the effectiveness of our approach through large-scale image generation experiments on ImageNet with diverse target distributions, accompanied by detailed discussions and ablation studies. With both theoretical and empirical validation, our framework paves a new way for latent space alignment.
format Preprint
id arxiv_https___arxiv_org_abs_2506_05240
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Aligning Latent Spaces with Flow Priors
Li, Yizhuo
Ge, Yuying
Ge, Yixiao
Shan, Ying
Luo, Ping
Machine Learning
Computer Vision and Pattern Recognition
This paper presents a novel framework for aligning learnable latent spaces to arbitrary target distributions by leveraging flow-based generative models as priors. Our method first pretrains a flow model on the target features to capture the underlying distribution. This fixed flow model subsequently regularizes the latent space via an alignment loss, which reformulates the flow matching objective to treat the latents as optimization targets. We formally prove that minimizing this alignment loss establishes a computationally tractable surrogate objective for maximizing a variational lower bound on the log-likelihood of latents under the target distribution. Notably, the proposed method eliminates computationally expensive likelihood evaluations and avoids ODE solving during optimization. As a proof of concept, we demonstrate in a controlled setting that the alignment loss landscape closely approximates the negative log-likelihood of the target distribution. We further validate the effectiveness of our approach through large-scale image generation experiments on ImageNet with diverse target distributions, accompanied by detailed discussions and ablation studies. With both theoretical and empirical validation, our framework paves a new way for latent space alignment.
title Aligning Latent Spaces with Flow Priors
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2506.05240