Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Deep Variational Implicit 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 |
|---|---|---|---|
| Stack implicit-process layers rather than use a single-layer VIP | VIP · Rectangles | accuracy (%): 85.63 ± 0.18 -> 88.21 ± 0.12 | Helped |
| Use a CNN prior in the first layer | DVIP · Rectangles | accuracy (%): 88.21 ± 0.12 vs 75.16 ± 0.16 | Helped |
| Increase depth from one layer to five layers | DVIP · SUSY | AUC: 0.8572 -> 0.8756 | Helped |
| Increase depth from one layer to five layers | DVIP · HIGGS | AUC: 0.6247 -> 0.7962 | Helped |
| Use DVIP rather than DGP | DGP · HIGGS | AUC: 0.8324 vs 0.7962 | Hurt |
| Increase prior sample count from 10 to 40 (ablation) | DVIP 3 · Power | RMSE: 4.03 ± 0.04 -> 3.92 ± 0.04 | Helped |
| Increase prior sample count from 10 to 50 (ablation) | DVIP 3 · Protein | RMSE: 4.53 ± 0.01 -> 4.21 ± 0.01 | Helped |
| Increase prior sample count from 20 to 50 (ablation) | DVIP 3 · Protein | CPU Time (s): 2734 ± 19 -> 4330 ± 52 | Hurt |
| Change prior BNN to 3 hidden layers of 10 units (ablation) | DVIP 3 · Protein | RMSE: 4.40 ± 0.01 -> 4.37 ± 0.01 | No clear effect |
| Change prior BNN to 2 hidden layers of 20 units (ablation) | DVIP 3 · Power | RMSE: 3.94 ± 0.04 -> 3.99 ± 0.04 | No clear effect |
| Replace BNN prior with approximate GP prior (ablation) | DVIP 3 · Protein | RMSE: 4.14 ± 0.01 -> 4.07 ± 0.01 | Helped |
| Use an unconstrained rather than constrained prior BNN (ablation) | VIP · Boston | RMSE Test: 4.73 ± 0.24 -> 5.85 ± 0.14 | Hurt |
Closest papers by meaning
- Function-space Inference with Sparse Implicit Processes (ICML 2022 · similarity 0.91)
- Variational Implicit Processes (ICML 2019 · similarity 0.91)
- Doubly Stochastic Variational Inference for Deep Gaussian Processes (NeurIPS 2017 · similarity 0.90)
- Implicit Posterior Variational Inference for Deep Gaussian Processes (NeurIPS 2019 · similarity 0.89)
- Sparse Gaussian Processes: Structured Approximations and Power-EP Revisited (NeurIPS 2025 · similarity 0.87)
- Sparse Inducing Points in Deep Gaussian Processes: Enhancing Modeling with Denoising Diffusion Variational Inference (ICML 2024 · similarity 0.86)
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.