Research map / Learning theory and algorithms / Prediction and decision losses
When Does Adaptation Win? Scaling Laws for Meta-Learning in Quantum Control
Nima Leclerc, Chris Miller, Nicholas Peter Brawand
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 |
|---|---|---|---|
| add K=10 task-specific adaptation steps | FOMAML meta-init · Two-qubit CZ gate, 10x OOD noise | gate fidelity: 54.2% -> 95.7% | Improved |
| add K=10 adaptation steps | FOMAML meta-init · Single-qubit X gate, mild OOD | fidelity F: 0.959 -> 0.960 | No clear effect |
| add K=10 adaptation steps | fixed-average baseline · Single-qubit X gate, mild OOD | fidelity F: 0.881 -> 0.935 | Improved |
| adaptation when task variance is small (sigma^2 < 0.002) | FOMAML · Single-qubit X gate | adaptation gap: G_inf ~ 0 | No clear effect |
| increase task variance | adaptation gap asymptote · Single-qubit X gate | asymptotic gap G_inf: linear scaling, R2 = 0.94 | Improved |
| add adaptation steps (K) beyond about 20 | FOMAML · Single-qubit X gate | adaptation gap: beta = 0.083 | No clear effect |
| adaptation alone without task parameter xi vs blind baseline (ablation) | FOMAML · Two-qubit gate, mild OOD | fidelity: +0.184 | Improved |
| adaptation alone without task parameter xi vs blind baseline (ablation) | FOMAML · Two-qubit gate, strong OOD 10x | fidelity: +0.158 | Improved |
| use 30% noisy xi instead of exact xi (ablation) | FOMAML · Two-qubit gate | fidelity | No clear effect |
| increase inner-loop learning rate beyond 2e-2 (ablation) | adaptation rate beta · Single-qubit X gate | adaptation rate beta | Hurt |
| increase inner-loop learning rate up to 2e-2 (ablation) | adaptation rate beta · Single-qubit X gate | adaptation rate beta | Improved |
| use meta-initialization instead of non-adaptive GRAPE | FOMAML (K=0) · Single-qubit X gate | gate fidelity (%): 97.9% -> 98.9% | Improved |
Closest papers by meaning
- Replay-buffer engineering for noise-robust quantum circuit optimization (NeurIPS 2026 · similarity 0.86)
- Reinforcement Learning for Quantum Control under Physical Constraints (ICML 2025 · similarity 0.86)
- Q-MAML: Quantum Model-Agnostic Meta-Learning for Variational Quantum Algorithms (AAAI 2025 · similarity 0.86)
- Curriculum reinforcement learning for quantum architecture search under hardware errors (ICLR 2024 · similarity 0.85)
- Differentiable Quantum Computing for Large-scale Linear Control (NeurIPS 2024 · similarity 0.85)
- Benchmarking Quantum Reinforcement Learning (ICML 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.
