Log-FM applies a per-coordinate soft-log transform (with Hill diagnostic) to compress heavy tails for standard flow matching, outperforming baselines on W1, CVaR99 and extreme quantiles across 144 multivariate configurations.
arXiv preprint arXiv:2410.14171 , year=
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
fields
stat.ML 3years
2026 3verdicts
UNVERDICTED 3representative citing papers
Proposes SDE sampler with state-dependent diffusion for HTDMs that induces self-regulating annealing, claimed necessary for heavy-tailed sampling.
Markov chain Phase-Type decoders in VAEs enable heavy-tailed generation where Gaussian decoders fail due to structural limitations from Lipschitz continuity.
citing papers explorer
-
Tail Annealing for Heavy-Tailed Flow Matching
Log-FM applies a per-coordinate soft-log transform (with Hill diagnostic) to compress heavy tails for standard flow matching, outperforming baselines on W1, CVaR99 and extreme quantiles across 144 multivariate configurations.
-
Self-Regulating Annealing in Heavy-Tailed Diffusion Models
Proposes SDE sampler with state-dependent diffusion for HTDMs that induces self-regulating annealing, claimed necessary for heavy-tailed sampling.
-
Markov Chain Decoders Overcome the Heavy-Tail Limitations of Lipschitz Generative Models
Markov chain Phase-Type decoders in VAEs enable heavy-tailed generation where Gaussian decoders fail due to structural limitations from Lipschitz continuity.