arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2408.15094cs.LGcs.CVmath.OC
收藏

基于对偶训练的约束扩散模型

Constrained Diffusion Models via Dual Training

University of Pennsylvania(宾夕法尼亚大学)

Shervin Khalafi, Dongsheng Ding, Alejandro Ribeiro

AI总结:

本文提出约束扩散模型,将需求约束下的训练建模为约束分布优化问题,并设计对偶训练算法,在公平采样和微调防过拟合任务中实现目标与约束的最优权衡。

摘要:

扩散模型因其能够通过扩散过程为给定数据集合成概率分布而备受关注,从而能够高保真地生成新数据点。然而,扩散过程容易生成反映训练数据集中偏差的样本。为解决这一问题,我们通过根据需求所指定的期望分布施加扩散约束,开发了约束扩散模型。具体而言,我们将需求约束下的扩散模型训练表述为一个约束分布优化问题,旨在减小原始数据与生成数据之间的分布差异,同时遵守对生成数据分布的约束。我们证明,所提出的约束扩散模型能够从混合数据分布中生成新数据,并在目标与约束之间实现最优权衡。为训练约束扩散模型,我们开发了一种对偶训练算法,并刻画了训练后的约束扩散模型的最优性。我们在两个约束生成任务中实证展示了约束模型的有效性:(i)考虑一个包含一个或多个代表性不足类别的数据集,训练时施加约束以确保推理阶段能够公平地从所有类别中采样;(ii)对预训练扩散模型进行微调,使其从新数据集中采样,同时避免过拟合。

英文摘要:

Diffusion models have attained prominence for their ability to synthesize a probability distribution for a given dataset via a diffusion process, enabling the generation of new data points with high fidelity. However, diffusion processes are prone to generating samples that reflect biases in a training dataset. To address this issue, we develop constrained diffusion models by imposing diffusion constraints based on desired distributions that are informed by requirements. Specifically, we cast the training of diffusion models under requirements as a constrained distribution optimization problem that aims to reduce the distribution difference between original and generated data while obeying constraints on the distribution of generated data. We show that our constrained diffusion models generate new data from a mixture data distribution that achieves the optimal trade-off among objective and constraints. To train constrained diffusion models, we develop a dual training algorithm and characterize the optimality of the trained constrained diffusion model. We empirically demonstrate the effectiveness of our constrained models in two constrained generation tasks: (i) we consider a dataset with one or more underrepresented classes where we train the model with constraints to ensure fairly sampling from all classes during inference; (ii) we fine-tune a pre-trained diffusion model to sample from a new dataset while avoiding overfitting.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑