A variational perspective on diffusion-based generative models and score matching

CW Huang, JH Lim… - Advances in Neural …, 2021‏ - proceedings.neurips.cc
Discrete-time diffusion-based generative models and score matching methods have shown
promising results in modeling high-dimensional image data. Recently, Song et al.(2021) …

Proximal langevin algorithm: Rapid convergence under isoperimetry

A Wibisono - arxiv preprint arxiv:1911.01469, 2019‏ - arxiv.org
We study the Proximal Langevin Algorithm (PLA) for sampling from a probability distribution
$\nu= e^{-f} $ on $\mathbb {R}^ n $ under isoperimetry. We prove a convergence guarantee …

Gaussian interpolation flows

Y Gao, J Huang, Y Jiao - arxiv preprint arxiv:2311.11475, 2023‏ - arxiv.org
Gaussian denoising has emerged as a powerful principle for constructing simulation-free
continuous normalizing flows for generative modeling. Despite their empirical successes …

Information geometry of dynamics on graphs and hypergraphs

TJ Kobayashi, D Loutchko, A Kamimura… - Information …, 2024‏ - Springer
We introduce a new information-geometric structure associated with the dynamics on
discrete objects such as graphs and hypergraphs. The presented setup consists of two …

Convexity of mutual information along the Fokker-Planck flow

J Zou, L Fan, J Gao, J Wang - arxiv preprint arxiv:2501.05094, 2025‏ - arxiv.org
We study the convexity of mutual information as a function of time along the Fokker-Planck
flow. The results are generalizations of that along heat flow and Ornstein-Ulenbeck flow …

Mixing Time of the Proximal Sampler in Relative Fisher Information via Strong Data Processing Inequality

A Wibisono - arxiv preprint arxiv:2502.05623, 2025‏ - arxiv.org
We study the mixing time guarantee for sampling in relative Fisher information via the
Proximal Sampler algorithm, which is an approximate proximal discretization of the …

McKean's conjecture under the log-concavity assumption

Y Geng - 2024 IEEE International Symposium on Information …, 2024‏ - ieeexplore.ieee.org
McKean conjectured that Gaussian random variables are optimal for the nth order derivative
of differential entropy along the heat flow, and verified this for n=1,2. Recently, Zhang …

On Independent Samples Along the Langevin Diffusion and the Unadjusted Langevin Algorithm

J Liang, S Mitra, A Wibisono - arxiv preprint arxiv:2402.17067, 2024‏ - arxiv.org
We study the rate at which the initial and current random variables become independent
along a Markov chain, focusing on the Langevin diffusion in continuous time and the …

Differential Properties of Information in Jump-diffusion Channels

L Fan, J Zou, J Wang - arxiv preprint arxiv:2501.05708, 2025‏ - arxiv.org
We propose a channel modeling using jump-diffusion processes, and study the differential
properties of entropy and mutual information. By utilizing the Kramers-Moyal and …

[كتاب][B] Information-based methods and models for particle flow filtering

KC Ward - 2021‏ - search.proquest.com
Recursive estimation methodologies, such as Kalman and Bayesian filters, typically require
models of some kind to perform the estimation. This filtering process seeks to improve …