TFG-Flow: Training-free Guidance in Multimodal Generative Flow

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Lin, Haowei, Li, Shanda, Ye, Haotian, Yang, Yiming, Ermon, Stefano, Liang, Yitao, Ma, Jianzhu
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912279225696256
author Lin, Haowei
Li, Shanda
Ye, Haotian
Yang, Yiming
Ermon, Stefano
Liang, Yitao
Ma, Jianzhu
author_facet Lin, Haowei
Li, Shanda
Ye, Haotian
Yang, Yiming
Ermon, Stefano
Liang, Yitao
Ma, Jianzhu
contents Given an unconditional generative model and a predictor for a target property (e.g., a classifier), the goal of training-free guidance is to generate samples with desirable target properties without additional training. As a highly efficient technique for steering generative models toward flexible outcomes, training-free guidance has gained increasing attention in diffusion models. However, existing methods only handle data in continuous spaces, while many scientific applications involve both continuous and discrete data (referred to as multimodality). Another emerging trend is the growing use of the simple and general flow matching framework in building generative foundation models, where guided generation remains under-explored. To address this, we introduce TFG-Flow, a novel training-free guidance method for multimodal generative flow. TFG-Flow addresses the curse-of-dimensionality while maintaining the property of unbiased sampling in guiding discrete variables. We validate TFG-Flow on four molecular design tasks and show that TFG-Flow has great potential in drug design by generating molecules with desired properties.
format Preprint
id arxiv_https___arxiv_org_abs_2501_14216
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle TFG-Flow: Training-free Guidance in Multimodal Generative Flow
Lin, Haowei
Li, Shanda
Ye, Haotian
Yang, Yiming
Ermon, Stefano
Liang, Yitao
Ma, Jianzhu
Machine Learning
Artificial Intelligence
Computational Engineering, Finance, and Science
Given an unconditional generative model and a predictor for a target property (e.g., a classifier), the goal of training-free guidance is to generate samples with desirable target properties without additional training. As a highly efficient technique for steering generative models toward flexible outcomes, training-free guidance has gained increasing attention in diffusion models. However, existing methods only handle data in continuous spaces, while many scientific applications involve both continuous and discrete data (referred to as multimodality). Another emerging trend is the growing use of the simple and general flow matching framework in building generative foundation models, where guided generation remains under-explored. To address this, we introduce TFG-Flow, a novel training-free guidance method for multimodal generative flow. TFG-Flow addresses the curse-of-dimensionality while maintaining the property of unbiased sampling in guiding discrete variables. We validate TFG-Flow on four molecular design tasks and show that TFG-Flow has great potential in drug design by generating molecules with desired properties.
title TFG-Flow: Training-free Guidance in Multimodal Generative Flow
topic Machine Learning
Artificial Intelligence
Computational Engineering, Finance, and Science
url https://arxiv.org/abs/2501.14216