Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Deep Bayesian Active Learning with Image Data
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 BALD acquisition instead of random sampling | Bayesian CNN · MNIST | acquired images to reach 5% error: 835 -> 335 | Helped |
| Use Variation Ratios acquisition instead of random sampling | Bayesian CNN · MNIST | acquired images to reach 5% error: 835 -> 295 | Helped |
| Use Max Entropy acquisition instead of random sampling | Bayesian CNN · MNIST | acquired images to reach 5% error: 835 -> 355 | Helped |
| Use Mean STD acquisition instead of random sampling | Bayesian CNN · MNIST | acquired images to reach 5% error: 695 vs 835 | No clear effect |
| Use Variation Ratios acquisition instead of random sampling | Bayesian CNN · MNIST, 1000 labelled training samples | test error: 4.66% -> 1.64% | Helped |
| Use Variation Ratios acquisition instead of random sampling | Bayesian CNN · ISIC 2016 melanoma diagnosis | acquisition usefulness | Hurt |
| Propagate parameter uncertainty using MC dropout rather than using a deterministic CNN | BALD, Variation Ratios, and Max Entropy acquisition · MNIST | test accuracy | Helped |
| Replace the RBF kernel with a CNN in MBR | MBR · MNIST, two-digit classification | classification accuracy | No clear effect |
| Use BALD instead of uniform acquisition | Fine-tuned VGG16 · ISIC 2016 melanoma diagnosis | AUC | Helped |
| Use BALD to select informative lesion images instead of uniform acquisition | Fine-tuned VGG16 · ISIC 2016 melanoma diagnosis | positive examples acquired | Helped |
Closest papers by meaning
- Bayesian Batch Active Learning as Sparse Subset Approximation (NeurIPS 2019 · similarity 0.87)
- Bayesian Generative Active Deep Learning (ICML 2019 · similarity 0.86)
- The Power of Ensembles for Active Learning in Image Classification (CVPR 2018 · similarity 0.86)
- Position: Bayesian Deep Learning is Needed in the Age of Large-Scale AI (ICML 2024 · similarity 0.85)
- Deep Active Learning with Noise Stability (AAAI 2024 · similarity 0.84)
- Think Twice Before Selection: Federated Evidential Active Learning for Medical Image Analysis with Domain Shifts (CVPR 2024 · similarity 0.83)
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.