Research map / Probabilistic and scientific modeling / Variational and generative sampling
Deep Stochastic Processes via Functional Markov Transition Operators
Jin Xu, Emilien Dupont, Kaspar Märtens, Tom Rainforth, Yee Whye Teh
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| MNPs (7 transition steps) instead of NPs | MNPs · 1D regression, RBF kernel | marginal log-likelihood: 0.935 -> 2.491 | Improved |
| MNPs instead of NPs | MNPs · 1D regression, Matern kernel | marginal log-likelihood: 1.115 -> 2.290 | Improved |
| MNPs instead of NPs | MNPs · 1D regression, periodic kernel | marginal log-likelihood: 0.356 -> 0.594 | Improved |
| MNPs instead of NPs on non-GP monotonic functions | MNPs · Monotonic functions | marginal log-likelihood: 1.823 -> 2.755 | Improved |
| MNPs instead of CBPs (Gaussian copula) on convex functions | MNPs · Convex functions | marginal log-likelihood: 4.268 -> 5.582 | Improved |
| MNPs instead of GPs on SDE samples | MNPs · SDEs | marginal log-likelihood: 1.719 -> 1.942 | Improved |
| MNPs instead of GPs/CBPs on GP-sampled RBF data | MNPs · RBF kernel | marginal log-likelihood: 2.844 -> 2.491 | Hurt |
| MNPs instead of CBPs on periodic kernel data | MNPs · Periodic kernel | marginal log-likelihood: 0.169 -> 0.594 | Improved |
| MNPs instead of NPs, cumulative regret at high exploration delta=0.99 | MNPs · Wheel bandit | cumulative regret: 22.13 -> 5.46 | Improved |
| MNPs instead of NPs, cumulative regret at delta=0.5 | MNPs · Wheel bandit | cumulative regret: 1.60 -> 1.08 | Improved |
| MNPs instead of NPs, simple regret at delta=0.5 | MNPs · Wheel bandit | simple regret: 1.04 -> 1.14 | No clear effect |
| MNPs instead of NPs, simple regret at delta=0.99 | MNPs · Wheel bandit | simple regret: 21.45 -> 5.22 | Improved |
Closest papers by meaning
- Rényi Neural Processes (ICML 2025 · similarity 0.86)
- Sequence Modeling with Spectral Mean Flows (NeurIPS 2025 · similarity 0.85)
- Stochastic Process Learning via Operator Flow Matching (NeurIPS 2025 · similarity 0.85)
- Efficient Training of Neural Stochastic Differential Equations by Matching Finite Dimensional Distributions (ICLR 2025 · similarity 0.85)
- Bridging discrete and continuous state spaces: Exploring the Ehrenfest process in time-continuous diffusion models (ICML 2024 · similarity 0.84)
- Foundation Inference Models for Markov Jump Processes (NeurIPS 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
