Research map / Architectures and efficiency / CNN design
Learning to Select Pre-Trained Deep Representations With Bayesian Evidence Framework
Check 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 Bayesian evidence to select and greedily combine pre-trained CNN representations rather than use a single representation | Transfer learning with pre-trained CNNs · 12 visual recognition datasets | test performance: 90.0 vs. 89.7 (PASCAL VOC 2007); 95.3 vs. 92.7 (Caltech 101); 94.5 vs. 91.3 (Oxford Flowers) | Helped |
| Replace grid-searched SVM regularization with Bayesian LS-SVM evidence maximization | CNN features for visual recognition · 12 visual recognition datasets | prediction accuracy: 85.8 vs. 85.8 (PASCAL VOC 2007, VGG19); 69.8 vs. 69.6 (Stanford 40 Action, GoogLeNet-ImageNet) | No clear effect |
| Use evidence-guided greedy ensemble selection instead of exhaustive oracle selection | Pre-trained CNN ensembles · Stanford 40 actions | mean AP: 77.8 vs. 78.6 | Hurt |
| Select CNN using maximum evidence rather than rely on a generic best-network assumption | Pre-trained CNN transfer learning · MIT Indoor Scene and SUN 397 Scene | test accuracy: 80.8 (GoogLeNet-Places) vs. 71.3 (GoogLeNet-ImageNet) on MIT; 59.8 vs. 48.3 on SUN | Mixed |
| Apply Aitken's delta-squared process to fixed-point evidence optimization | Bayesian LS-SVM · PASCAL VOC 2012, aeroplane class | convergence speed | Helped |
Closest papers by meaning
- Bayesian Low-Rank Learning (Bella): A Practical Approach to Bayesian Neural Networks (AAAI 2025 · similarity 0.86)
- Bayesian Exploration of Pre-trained Models for Low-shot Image Classification (CVPR 2024 · similarity 0.85)
- A Bayesian Approach to Data Point Selection (NeurIPS 2024 · similarity 0.85)
- Knowledge Transfer from Vision Foundation Models for Efficient Training of Small Task-specific Models (ICML 2024 · similarity 0.84)
- Variational Bayesian Pseudo-Coreset (ICLR 2025 · similarity 0.84)
- Scalable Bayesian Learning with posteriors (ICLR 2025 · 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.