Research map / Probabilistic and scientific modeling / Variational and generative sampling
Sliced Kernelized Stein Discrepancy
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use maxSKSD instead of KSD for ICA training | ICA · ICA, D=200 | test NLL: -329.49 -> 47.72 | Helped |
| Use maxSKSD instead of LSD for ICA training | ICA · ICA, D=200 | test NLL: 46.63 -> 47.72 | Helped |
| Use S-SVGD instead of SVGD | VAE · MNIST, latent D=16 | test log-likelihood: -88.58 -> -89.17 | Hurt |
| Use S-SVGD instead of SVGD | VAE · MNIST, latent D=32 | test log-likelihood: -90.43 -> -87.55 | Helped |
| Use S-SVGD instead of SVGD | VAE · MNIST, latent D=48 | test log-likelihood: -93.47 -> -87.74 | Helped |
| Use S-SVGD instead of SVGD | VAE · MNIST, latent D=64 | test log-likelihood: -94.88 -> -87.78 | Helped |
| Use S-SVGD instead of SVGD | VAE imputation · MNIST imputation | label entropy: 0.538 -> 0.542 | Helped |
| Use S-SVGD instead of SVGD | VAE imputation · MNIST imputation | accuracy: 0.691 -> 0.728 | Helped |
| Optimize slicing direction r as well as g (maxSKSD-rg rather than maxSKSD-g) | RBM goodness-of-fit test · 50-dim RBM, perturbation level 0.01 | rejection rate: 0.96; others are all below 0.5 | Helped |
| Use one-dimensional projected kernels and test functions (maxSKSD) rather than high-dimensional KSD | Goodness-of-fit tests · High-dimensional Gaussian benchmarks | rejection rate | Helped |
| Use one-dimensional projected kernels in S-SVGD rather than SVGD's kernel | Particle inference · High-dimensional Gaussian variance estimation | variance estimation robustness | Helped |
Closest papers by meaning
- Active Slices for Sliced Stein Discrepancy (ICML 2021 · similarity 0.92)
- Kernel Stein Discrepancy Descent (ICML 2021 · similarity 0.91)
- Stein Variational Gradient Descent as Moment Matching (NeurIPS 2018 · similarity 0.89)
- The Polynomial Stein Discrepancy for Assessing Moment Convergence (ICML 2025 · similarity 0.88)
- ELBOing Stein: Variational Bayes with Stein Mixture Inference (ICLR 2025 · similarity 0.87)
- Generative Sliced MMD Flows with Riesz Kernels (ICLR 2024 · similarity 0.87)
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.