5 papers
Noise is All You Need: Solving Linear Inverse Problems by Noise Combination Sampling with Diffusion Models
Xun Su, Hiroyuki Kasai
Pretrained diffusion models have demonstrated strong capabilities in zero-shot inverse problem solving by incorporating observation information into the generation process of the d…
Navigating the Exploration-Exploitation Tradeoff in Inference-Time Scaling of Diffusion Models
Xun Su, Jianming Huang, Yang Yusen +2
Inference-time scaling has achieved remarkable success in language models, yet its adaptation to diffusion models remains underexplored. We observe that the efficacy of recent Sequ…
Anchor Space Optimal Transport as a Fast Solution to Multiple Optimal Transport Problems
Jianming Huang, Xun Su, Zhongxi Fang +1
In machine learning, Optimal Transport (OT) theory is extensively utilized to compare probability distributions across various applications, such as graph data represented by node…
Safe Screening for Unbalanced Optimal Transport
Xun Su, Zhongxi Fang, Hiroyuki Kasai
This paper introduces a framework that utilizes the Safe Screening technique to accelerate the optimization process of the Unbalanced Optimal Transport (UOT) problem by proactively…
Wasserstein Graph Distance Based on -Approximated Tree Edit Distance between Weisfeiler-Lehman Subtrees
Zhongxi Fang, Jianming Huang, Xun Su +1
The Weisfeiler-Lehman (WL) test is a widely used algorithm in graph machine learning, including graph kernels, graph metrics, and graph neural networks. However, it focuses only on…