Research map / Optimization / Nonconvex and smooth optimization
Biased Stochastic First-Order Methods for Conditional Stochastic Optimization and Applications in Meta 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 BSGD with small inner batch instead of SAA for the same sample budget | invariant logistic regression · noise ratio 1 | error (mean): 5.56e-04 -> 1.77e-04 | Helped |
| use BSGD instead of SAA | invariant logistic regression · noise ratio 100 | error (mean): 1.12e-02 -> 1.50e-03 | Helped |
| use BSGD (Hessian-aware biased gradient) instead of first-order MAML | sine-wave MAML · m=20 | average loss: 2.50e+00 -> 2.04e-01 | Helped |
| use BSGD instead of FO-MAML | sine-wave MAML · m=10 | CPU minutes: 41.45 -> 71.57 | Hurt |
| use BSGD instead of Adam on biased gradient | sine-wave MAML · m=10 | average loss: 8.16e-01 -> 2.12e-01 | Helped |
| increase inner batch size m from 10 to 50 for BSGD (ablation) | BSGD · sine-wave MAML | average loss: 2.12e-01 -> 2.17e-01 | No clear effect |
| use BSpiderBoost instead of BSGD | sine-wave MAML · convergence and recovery | convergence | Helped |
Closest papers by meaning
- A Guide Through the Zoo of Biased SGD (NeurIPS 2023 · similarity 0.89)
- Generalization Analysis of Biased Stochastic Gradient Methods for Minimax Problems (NeurIPS 2026 · similarity 0.88)
- Non-asymptotic Analysis of Biased Adaptive Stochastic Approximation (NeurIPS 2024 · similarity 0.87)
- Clipped SGD Algorithms for Performative Prediction: Tight Bounds for Stochastic Bias and Remedies (ICML 2025 · similarity 0.86)
- Distributionally Robust Optimization with Bias and Variance Reduction (ICLR 2024 · similarity 0.86)
- Stochastic Optimization Schemes for Performative Prediction with Nonconvex Loss (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.