Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Doubly Stochastic Variational Inference for Deep Gaussian Processes
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 a 4-layer DGP instead of a 2-layer DGP (ablation) | Doubly stochastic DGP · Rectangles-Images | accuracy (%): 77.3 -> 77.9 | Helped |
| Increase depth from 2 to 5 layers (ablation) | Doubly stochastic DGP · year | RMSE: 9.58 -> 8.87 | Helped |
| Increase depth from 2 to 5 layers (ablation) | Doubly stochastic DGP · taxi | RMSE: 281.4 -> 266.4 | Helped |
| Use a deep GP instead of a 500-inducing-point sparse GP | Doubly stochastic DGP · taxi | RMSE: 330.7 -> 266.4 | Helped |
| Increase depth from 2 to 3 layers (ablation) | Doubly stochastic DGP · MNIST | accuracy (%): 98.06% -> 98.11% | Helped |
| Increase depth from 2 to 5 layers (ablation) | Doubly stochastic DGP · HIGGS | AUC: 0.830 -> 0.846 | Helped |
| Increase depth beyond 2 layers (ablation) | Doubly stochastic DGP · SUSY | AUC: 0.877 -> 0.877 | No clear effect |
| Use a DGP instead of a 500-inducing-point sparse GP | Doubly stochastic DGP · kin8nm | test log likelihood: 1.15(0.00) -> 1.38(0.01) | Helped |
| Use a DGP instead of a 500-inducing-point sparse GP | Doubly stochastic DGP · protein | test log likelihood: -2.83(0.00) -> -2.71(0.00) | Helped |
Closest papers by meaning
- Inference in Deep Gaussian Processes using Stochastic Gradient Hamiltonian Monte Carlo (NeurIPS 2018 · similarity 0.93)
- Deep Gaussian Processes with Importance-Weighted Variational Inference (ICML 2019 · similarity 0.91)
- Random Feature Expansions for Deep Gaussian Processes (ICML 2017 · similarity 0.91)
- Deep Variational Implicit Processes (ICLR 2023 · similarity 0.90)
- Variational Gaussian Processes with Decoupled Conditionals (NeurIPS 2023 · similarity 0.90)
- Sparse Inducing Points in Deep Gaussian Processes: Enhancing Modeling with Denoising Diffusion Variational Inference (ICML 2024 · 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.