Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Advanced Dropout: A Model-free Methodology for Bayesian Dropout Optimization
Jiyang Xie, Zhanyu Ma, Jianjun Lei, Guoqiang Zhang, Jing‐Hao Xue, Zheng‐Hua Tan, Jun Guo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add SGVB optimization of mask distribution parameters (vs fixed at 0.5) (ablation) | advanced dropout · CIFAR-10 (VGG16) | accuracy (%): 93.98 -> 94.28 | Improved |
| add parametric prior (ablation) | advanced dropout · CIFAR-10 (VGG16) | accuracy (%): 94.20 -> 94.28 | Improved |
| add SGVB optimization of mask distribution parameters (ablation) | advanced dropout · CIFAR-100 (VGG16) | accuracy (%): 74.55 -> 74.94 | Improved |
| add parametric prior (ablation) | advanced dropout · CIFAR-100 (VGG16) | accuracy (%): 74.77 -> 74.94 | Improved |
| add SGVB optimization of mask distribution parameters (ablation) | advanced dropout · MNIST | accuracy (%): 98.70 -> 98.89 | Improved |
| add parametric prior (ablation) | advanced dropout · MNIST | accuracy (%): 98.81 -> 98.89 | Improved |
| add parametric prior (ablation) | advanced dropout · miniImageNet (VGG16) | accuracy (%): 77.30 -> 77.35 | Improved |
| add parametric prior (ablation) | advanced dropout · Caltech-256 (VGG16) | accuracy (%): 65.03 -> 65.09 | Improved |
| apply advanced dropout to base model | VGG16 · ImageNet (top-1) | accuracy (%): 73.42 -> 74.82 | Improved |
| apply advanced dropout to base model | VGG16 · CIFAR-100 | accuracy (%): 73.62 -> 74.94 | Improved |
| apply advanced dropout to base model | WRN16-10 · CIFAR-10 | accuracy (%): 94.29 -> 95.50 | Improved |
| apply advanced dropout to base model | DenseNet40 · CIFAR-100 | accuracy (%): 68.36 -> 70.65 | Improved |
Closest papers by meaning
- Dropout Reduces Underfitting (ICML 2023 · similarity 0.92)
- Dropout with Expectation-linear Regularization (ICLR 2017 · similarity 0.92)
- Contextual Dropout: An Efficient Sample-Dependent Dropout Module (ICLR 2021 · similarity 0.92)
- Demystifying Dropout (ICML 2019 · similarity 0.92)
- Jumpout : Improved Dropout for Deep Neural Networks with ReLUs (ICML 2019 · similarity 0.91)
- Stochastic Modified Equations and Dynamics of Dropout Algorithm (ICLR 2024 · similarity 0.89)
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.
