Research map / Learning theory and algorithms / Prediction and decision losses
Minimizing the Maximal Loss: How and Why
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 focused example sampling instead of uniform SGD sampling | SGD on the LeNet variant CNN · Face detection train | error: 0.1313% -> zero training error after 30 epochs | Helped |
| Use focused example sampling instead of uniform SGD sampling | SGD on the LeNet variant CNN · Face detection test | error: 0.35% -> 0.14% | Helped |
| Use FOL rather than AdaBoost with SGD as weak learner | LeNet variant CNN · Face detection train | epochs to consistent hypothesis: 20 epochs -> 27 epochs | Hurt |
| Use a single final FOL hypothesis instead of AdaBoost's weighted ensemble at prediction time | Face detection predictor · Face detection inference | relative prediction time: 10 times slower | Helped |
| Focus updates on currently informative examples | SGD on the face detection training set · Face detection | informative samples per iteration: SGD picks one useful sample only once every fifteen iterations; FOL expects an average of 32 useful samples in every ite | Helped |
Closest papers by meaning
- A Unified Framework for Rank-based Loss Minimization (NeurIPS 2023 · similarity 0.87)
- Penalizing Gradient Norm for Efficiently Improving Generalization in Deep Learning (ICML 2022 · similarity 0.87)
- A Minimax Approach to Supervised Learning (NeurIPS 2016 · similarity 0.86)
- From Average Sensitivity to Small-Loss Regret Bounds under Random-Order Model (NeurIPS 2026 · similarity 0.85)
- Thumb on the Scale: Optimal Loss Weighting in Last Layer Retraining (NeurIPS 2025 · similarity 0.84)
- Newton Losses: Using Curvature Information for Learning with Differentiable Algorithms (NeurIPS 2024 · similarity 0.84)
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.