Research map / Optimization / Nonconvex and smooth optimization
Adaptive Sampling for Stochastic Risk-Averse 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 adaptive DPP-based sampling (Ada-CVaR) instead of truncated-loss CVaR (Trunc-CVaR) | CVaR learner · classification (UCI, logistic regression) | accuracy: 0.71 -> 0.85 (Adult) | Helped |
| use Ada-CVaR instead of Mean (ERM) | logistic regression · UCI classification | CVaR of surrogate loss | Helped |
| use Ada-CVaR instead of Mean (ERM) | linear regression · Abalone, Boston, Cpu | CVaR of loss: 11.07 -> 8.92 (Abalone) | Helped |
| use Ada-CVaR instead of Trunc-CVaR | linear regression · Abalone, Boston, Cpu | CVaR of loss: 7.94 vs 8.92 (Abalone) | No clear effect |
| use Ada-CVaR instead of ERM/Trunc/Soft | classifiers under 10%/90% class imbalance shift · UCI classification | test accuracy | Helped |
| use Ada-CVaR instead of ERM, Soft, Trunc | VGG16-BN and ResNet-18 · CIFAR-10 with power-law class shift | test accuracy | Helped |
| use Ada-CVaR instead of Trunc-CVaR on CIFAR-10 | deep network · CIFAR-10 | CVaR of surrogate loss | Hurt |
| use Ada-CVaR instead of Soft-CVaR for training speed | deep network · training speed | epochs to 85% train accuracy: 21 -> 7 | Helped |
| use Ada-CVaR instead of Mean (ERM) | ResNet-18 · CIFAR-10 with distribution shift | test accuracy | No clear effect |
Closest papers by meaning
- Risk-Averse No-Regret Learning in Online Convex Games (ICML 2022 · similarity 0.91)
- Distributionally-Aware Kernelized Bandit Problems for Risk Aversion (ICML 2022 · similarity 0.89)
- Fast and Robust: Task Sampling with Posterior and Diversity Synergies for Adaptive Decision-Makers in Randomized Environments (ICML 2025 · similarity 0.88)
- Adaptive Regret for Bandits Made Possible: Two Queries Suffice (ICLR 2024 · similarity 0.86)
- Distributionally Robust Optimization with Bias and Variance Reduction (ICLR 2024 · similarity 0.86)
- When Lower-Order Terms Dominate: Adaptive Expert Algorithms for Heavy-Tailed Losses (NeurIPS 2025 · 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.