Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Structured and Efficient Variational Deep Learning with Matrix Gaussian Posteriors
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 |
|---|---|---|---|
| matrix variate Gaussian posterior (VMG) instead of fully factorized posterior (VI) | Bayesian neural network regression · UCI regression (10 datasets) | Avg. test RMSE | Helped |
| matrix variate Gaussian posterior with pseudo-data and GP interpretation | MNIST permutation invariant with small networks · MNIST test error | Test error (%): 3x150: 1.18 vs variational dropout ≈1.42 | Helped |
| increasing network width (VMG 3x150 to 3x750) | VMG on MNIST · MNIST test error | Test error (%): 1.18 -> 1.05 | Helped |
| matrix variate Gaussian posterior vs PBP and Dropout on Yacht/Boston/Energy | UCI Yacht RMSE · UCI Yacht | Avg. test RMSE | Mixed |
| pseudo-data (10 pairs per layer) with matrix variate posterior | UCI regression · UCI regression (Table 1) | Avg. test RMSE | Mixed |
| fully factorized Gaussian posterior with local reparametrization (no matrix structure) on toy data (ablation) | toy 1D regression · toy dataset | qualitative fit | Hurt |
Closest papers by meaning
- The k-tied Normal Distribution: A Compact Parameterization of Gaussian Mean Field Posteriors in Bayesian Neural Networks (ICML 2020 · similarity 0.92)
- Beyond the Mean-Field: Structured Deep Gaussian Processes Improve the Predictive Uncertainties (NeurIPS 2020 · similarity 0.92)
- Efficient Low Rank Gaussian Variational Inference for Neural Networks (NeurIPS 2020 · similarity 0.91)
- VIKING: Deep variational inference with stochastic projections (NeurIPS 2025 · similarity 0.89)
- Amortized Structured Stochastic Variational Inference for Gaussian Process Latent Variable Models (NeurIPS 2026 · similarity 0.88)
- Variational Inference with Mixtures of Isotropic Gaussians (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.