Research map / Learning theory and algorithms / Prediction and decision losses
A Lagrangian Duality Approach to Active 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 ALLY with k-means batch diversity instead of the best baseline | Active learning · STL-10, budget 200 | samples needed to reach 97% of final accuracy: 9% | Helped |
| Use ALLY with k-means batch diversity instead of the best baseline | Active learning · SVHN, budget 1000 | samples needed to reach 97% of final accuracy: 8% | Helped |
| Use ALLY with k-means batch diversity instead of the best baseline | Active learning · CIFAR-10, budget 200 | samples needed to reach 97% of final accuracy: 2% | Helped |
| Use ALLY instead of BADGE or BAIT | Active learning with an MLP · MNIST | test accuracy | Hurt |
| Use ALLY instead of Random or Coreset | Active learning · Parkinsons Telemonitoring dataset | mean-squared error | Helped |
| Use ALLY instead of Entropy Sampling (ablation) | Active learning with redundant training data · STL-10 | test accuracy | Helped |
| Query multiple samples per batch rather than one sample per cluster (ablation) | ALLY · STL-10 with dataset redundancy | test accuracy | Hurt |
| Increase the number of k-means clusters (ablation) | ALLY · MNIST | test accuracy | Helped |
| Set constraint tightness extremely high, such as 20 nats (ablation) | ALLY · STL-10 and MNIST | informativeness of dual-variable scores: 20 nats | Hurt |
| Set constraint tightness to a large value, such as 2 nats (ablation) | ALLY · STL-10 and MNIST | outlier detection: 2 nats | Helped |
Closest papers by meaning
- Near-Optimal Solutions of Constrained Learning Problems (ICLR 2024 · similarity 0.87)
- Neural Active Learning Beyond Bandits (ICLR 2024 · similarity 0.87)
- Active Classification with Few Queries under Misspecification (NeurIPS 2024 · similarity 0.86)
- Contextual Active Model Selection (NeurIPS 2024 · similarity 0.86)
- Active Learning of General Halfspaces: Label Queries vs Membership Queries (NeurIPS 2024 · similarity 0.86)
- Transductive Active Learning: Theory and Applications (NeurIPS 2024 · similarity 0.86)
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.