Research map / Architectures and efficiency / RNNs and normalization
Excitation Dropout: Encouraging Plasticity in Deep Neural Networks
Andrea Zunino, Sarah Adel Bargal, Pietro Morerio, Jianming Zhang, Stan Sclaroff, Vittorio Murino
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 excitation-based retaining probabilities (vs standard dropout, both fcs) | Dropout in fc layers · UCF101 AlexNet | test accuracy (%): 65.83 -> 67.56 | Improved |
| use input-specific masks guided by excitation rather than random masks per image (ablation) | Standard Dropout · Cifar10/Cifar100/Caltech256 | test accuracy | Improved |
| apply Excitation Dropout at conv3 instead of fc1 (ablation) | Excitation Dropout · Cifar10 (CNN-2) | test accuracy (%): 81.94 -> 78.01 | Hurt |
| apply Excitation Dropout at conv3 instead of no dropout (ablation) | CNN-2 · Cifar10 | test accuracy (%): 76.91 -> 78.01 | Improved |
| replace Excitation Dropout by Activation Dropout (ablation) | Excitation Dropout · Cifar10 | test accuracy (%): 81.94 -> 81.33 | Hurt |
| replace Excitation Dropout by Activation Dropout (ablation) | Excitation Dropout · Caltech256 | test accuracy (%): 35.77 -> 35.29 | Hurt |
| replace Excitation Dropout by Activation Dropout (ablation) | Excitation Dropout · UCF101 VGG16 | test accuracy (%): 73.23 -> 72.56 | Hurt |
| use Activation Dropout instead of Standard Dropout on WRN-28-10 | WideResNet-28-10 · Cifar10 | test error (%): 4.17 -> 3.88 | Improved |
| use Activation Dropout instead of Standard Dropout on AlexNet | AlexNet from scratch · ImageNet | test accuracy (%): 44.43 -> 48.54 | Improved |
| add Excitation Dropout instead of Standard Dropout in training time | Standard Dropout · VGG16 training | time per 100 iterations (s): 2.279 -> 2.820 | Hurt |
| increase dropout rate from 0.25 to 0.75 for Excitation Dropout (ablation) | Excitation Dropout · Cifar10 | test accuracy (%): 81.38 -> 81.55 | Mixed |
| increase dropout rate from 0.25 to 0.75 for Standard Dropout (both fcs) (ablation) | Standard Dropout · Cifar100 | test accuracy (%): 50.30 -> 49.46 | Mixed |
Closest papers by meaning
- Dropout with Expectation-linear Regularization (ICLR 2017 · similarity 0.89)
- Jumpout : Improved Dropout for Deep Neural Networks with ReLUs (ICML 2019 · similarity 0.89)
- Improved Dropout for Shallow and Deep Learning (NeurIPS 2016 · similarity 0.88)
- Guided Dropout (AAAI 2019 · similarity 0.88)
- Activation by Interval-wise Dropout: A Simple Way to Prevent Neural Networks from Plasticity Loss (ICML 2025 · similarity 0.88)
- Information Dropout: Learning Optimal Representations Through Noisy Computation (TPAMI 2018 · 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.
