Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning
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 MC dropout uncertainty | Dropout neural networks · Boston Housing, Concrete Strength, Energy Efficiency, Kin8nm, Naval Propulsion, Power Plant, Protein Structure, Wine Quality Red, Yacht Hydrodynamics | predictive log-likelihood: Boston Housing: -2.57 $\pm$ 0.09 -> -2.46 $\pm$ 0.06; Concrete Strength: -3.16 $\pm$ 0.02 -> -3.04 $\pm$ 0.02; Energy Ef | Helped |
| Use MC dropout uncertainty | Dropout neural networks · Boston Housing, Concrete Strength, Energy Efficiency, Power Plant, Protein Structure, Wine Quality Red | RMSE: Boston Housing: 3.01 $\pm$ 0.18 -> 2.97 $\pm$ 0.19; Concrete Strength: 5.67 $\pm$ 0.09 -> 5.23 $\pm$ 0.12; Energy Effici | Helped |
| Use MC dropout uncertainty | Dropout neural networks · Kin8nm and Naval Propulsion | RMSE: Kin8nm: 0.10 $\pm$ 0.00 -> 0.10 $\pm$ 0.00; Naval Propulsion: 0.01 $\pm$ 0.00 -> 0.01 $\pm$ 0.00 | No clear effect |
| Use MC dropout uncertainty | Dropout neural networks · Yacht Hydrodynamics | RMSE: 1.02 $\pm$ 0.05 -> 1.11 $\pm$ 0.09 | Hurt |
| Train for 10 times the number of epochs (ablation) | Dropout on Boston Housing · Boston Housing | RMSE: 2.97 $\pm$ 0.19 -> 2.80 $\pm$ 0.19 | Helped |
| Train for 10 times the number of epochs (ablation) | Dropout on Boston Housing · Boston Housing | predictive log-likelihood: -2.46 $\pm$ 0.06 -> -2.39 $\pm$ 0.05 | Helped |
| Use 2 layers instead of 1 (ablation) | Dropout on Boston Housing · Boston Housing | RMSE: 2.97 $\pm$ 0.19 -> 2.80 $\pm$ 0.13 | Helped |
| Use 2 layers instead of 1 (ablation) | Dropout on Boston Housing · Boston Housing | predictive log-likelihood: -2.46 $\pm$ 0.06 -> -2.34 $\pm$ 0.02 | Helped |
| Set dropout probability to 0 (ablation) | Dropout on Boston Housing · Boston Housing | RMSE: 2.97 -> 3.07 | Hurt |
| Set dropout probability to 0 (ablation) | Dropout on Boston Housing · Boston Housing | predictive log-likelihood: -2.46 -> -2.59 | Hurt |
Closest papers by meaning
- Concrete Dropout (NeurIPS 2017 · similarity 0.92)
- Dropout Inference in Bayesian Neural Networks with Alpha-divergences (ICML 2017 · similarity 0.91)
- Variational Dropout Sparsifies Deep Neural Networks (ICML 2017 · similarity 0.90)
- A Theoretically Grounded Application of Dropout in Recurrent Neural Networks (NeurIPS 2016 · similarity 0.90)
- Uncertainty Quantification for Deep Regression using Contextualised Normalizing Flows (NeurIPS 2025 · similarity 0.87)
- Stochastic Modified Equations and Dynamics of Dropout Algorithm (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.