Research map / Probabilistic and scientific modeling / Variational and generative sampling
Kernel Implicit Variational Inference
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 KIVI instead of factorized Gaussian inference | BNN · Boston, Concrete, Energy, Kin8nm, Naval, Combined, Protein, Yacht | RMSE | Helped |
| Use KIVI instead of SVGD | BNN · Wine and Year | RMSE | Hurt |
| Use KIVI instead of a plain VAE | VAE · Statically binarized MNIST | test log-likelihood: -97.3 -> -94.8 | Helped |
| Use KIVI instead of AVB | VAE · CelebA | FID: 160 -> 41 | Helped |
| Add the reverse ratio trick, estimating p/q rather than q/p (ablation) | KIVI · 2-D Bayesian logistic regression | posterior approximation quality | Helped |
| Replace the factorized Gaussian posterior with an implicit posterior | VI on a 1-D Gaussian mixture · 1-D Gaussian mixture | mode coverage | Helped |
| Replace factorized VI with planar normalizing flows | BNN regression · Boston, Concrete, Energy, Kin8nm, Naval | RMSE | No clear effect |
| Use KIVI rather than prior-contrastive logistic regression | BNN classification · MNIST | training lower bound | Helped |
| Add implicit posterior averaging across a broader range of configurations | VAE · Statically binarized MNIST | train-test lower-bound gap | Helped |
| Increase the discrepancy between the contrast distribution and the variational posterior | KIVI density-ratio estimation · Local latent variable models | density-ratio estimation accuracy | Hurt |
Closest papers by meaning
- Kernel Semi-Implicit Variational Inference (ICML 2024 · similarity 0.92)
- Implicit Variational Inference for High-Dimensional Posteriors (NeurIPS 2023 · similarity 0.91)
- Globally Convergent Variational Inference (NeurIPS 2024 · similarity 0.89)
- Revisiting Unbiased Implicit Variational Inference (ICML 2025 · similarity 0.89)
- Variational Inference with Mixtures of Isotropic Gaussians (NeurIPS 2025 · similarity 0.88)
- VIKING: Deep variational inference with stochastic projections (NeurIPS 2025 · similarity 0.88)
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.