From Global to Tail Convergence: Dynamics of alpha-Stable Score-Based Generative Models
cs.LG, math.PR, stat.ML
Submitted: 2026-10-14
Updated: 2026-10-14
Comments: 44 pages, 4 figures, 3 tables. Code: https://github.com/ainsi333/stable-sbgm-experiments
Code: https://github.com/ainsi333/stable-sbgm-experiments
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Heavy-Tailed Diffusion Models
- Generalizing Score-based generative models for Heavy-tailed Distributions
- Non Gaussian Denoising Diffusion Models
- Heavy-tailed denoising score matching
- Heavy-Tailed Diffusion with Denoising L'evy Probabilistic Models
- Do Heavy Tails Help Diffusion? On the Subtle Trade-off Between Initialization and Training
- Generator-Guided Inverse Sampling for L'evy-Driven Generative Models
- Superposition principle for non-local Fokker-Planck operators
- Density and tails of unimodal convolution semigroups
- Estimates of transition densities and their derivatives for jump L'evy processes
- Extremal behavior of stochastic integrals driven by regularly varying L'{e}vy processes
- Time reversal of diffusion processes under a finite entropy condition
- An Introduction to L'evy and Feller Processes. Advanced Courses in Mathematics - CRM Barcelona 2014
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks