Research map / Optimization / Optimizers
BiSLS/SPS: Auto-tune Step Sizes for Stable Bi-level Optimization
Chen Fan, Gaspard Choné-Ducasse, Mark Schmidt, Christos Thrampoulidis
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 reset option 3 with eta=2 instead of reset option 1 (ablation) | BiSLS-SGD · Hyper-representation MNIST | search cost per iteration: 89 -> 9 | Improved |
| use reset option 3 with eta=2 instead of reset option 1 (ablation) | BiSLS-Adam · Hyper-representation MNIST | search cost per iteration: 115 -> 9 | Improved |
| use eta=1 (reset option 2) instead of eta=2 (ablation) | BiSLS · Hyper-representation MNIST | search cost per iteration: 9 -> 4 | Mixed |
| add line search at the upper level (ablation) | bi-level SGD/Adam · Hyper-representation MNIST | validation loss | Improved |
| use a single gradient step to approximate y*(x) (ablation) | BiSLS · Hyper-representation MNIST | validation accuracy vs run time | No clear effect |
| replace fine-tuned Adam with BiSLS-Adam | Adam · Hyper-representation MNIST | time to 85% validation accuracy | Improved |
| vary initial search step sizes alpha_b0 and beta_b0 (ablation) | BiSLS-Adam/SGD · Hyper-representation MNIST | validation loss | No clear effect |
| replace fine-tuned Adam/SGD with BiSLS-SGD | SGD/Adam · Data distillation MNIST | validation loss convergence | Improved |
| use one sample versus two samples for upper-level step size (ablation) | BiSPS · bi-level | empirical performance | No clear effect |
Closest papers by meaning
- A Stochastic Approach to Bi-Level Optimization for Hyperparameter Optimization and Meta Learning (AAAI 2025 · similarity 0.87)
- Memory-Efficient Gradient Unrolling for Large-Scale Bi-level Optimization (NeurIPS 2024 · similarity 0.87)
- Moreau Envelope for Nonconvex Bi-Level Optimization: A Single-Loop and Hessian-Free Solution Strategy (ICML 2024 · similarity 0.86)
- Adaptive Methods through the Lens of SDEs: Theoretical Insights on the Role of Noise (ICLR 2025 · similarity 0.85)
- BROS: Bias-Corrected Randomized Subspaces for Memory-Efficient Single-Loop Bilevel Optimization (NeurIPS 2026 · similarity 0.85)
- Enhancing Optimizer Stability: Momentum Adaptation of The NGN Step-size (NeurIPS 2025 · similarity 0.85)
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.
