Research map / Architectures and efficiency / Neural architecture search
Multi-Fidelity Automatic Hyper-Parameter Tuning via Transfer Series Expansion
Yi-Qi Hu, Yang Yu, Wei-Wei Tu, Qiang Yang, Yuqiang Chen, Wenyuan Dai
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 TSE-corrected high-fidelity residual predictor to low-fidelity-only optimisation (ablation) | SRACOS on LightGBM · Connect | test AUC: 0.9272 -> 0.9374 | Improved |
| add TSE-corrected low-fidelity evaluations (ablation) | SRACOS on LightGBM · Airline | test AUC: 0.8693 -> 0.8893 | Improved |
| add TSE-corrected low-fidelity evaluations (ablation) | SRACOS on LightGBM · Criteo | test AUC: 0.7480 -> 0.7496 | Improved |
| add TSE correction (ablation) | SRACOS on LightGBM · Credit | test AUC: 0.7531 -> 0.7612 | Improved |
| replace random forest residual predictor with linear-combination TSE (ablation) | TSESRACOS · Connect | test AUC: 0.9330 -> 0.9374 | Improved |
| replace random forest residual predictor with TSE (ablation) | TSESRACOS · Airline | test AUC: 0.8762 -> 0.8893 | Improved |
| replace random forest residual predictor with TSE (ablation) | TSESRACOS · Musk | test AUC: 0.9980 -> 0.9977 | Mixed |
| transfer base predictors from Miniboone instead of training own base predictors (ablation) | TSESRACOS · Higgs | test AUC: 0.8023 -> 0.8044 | No clear effect |
| multi-fidelity TSE vs full high-fidelity optimisation | TSESRACOS · Airline wall-clock | test AUC: 0.8961 (HF-ONLY*, 2:00:00) vs 0.8893 (0:44:06) | Mixed |
| TSE linear combination vs random forest regression error (ablation) | TSESRACOS · 8 datasets regression error | absolute regression error | Improved |
Closest papers by meaning
- Multi-Fidelity Black-Box Optimization with Hierarchical Partitions (ICML 2018 · similarity 0.86)
- An ADMM Based Framework for AutoML Pipeline Configuration (AAAI 2020 · similarity 0.86)
- AUTOMATA: Gradient Based Data Subset Selection for Compute-Efficient Hyper-parameter Tuning (NeurIPS 2022 · similarity 0.85)
- Weighted Sampling for Combined Model Selection and Hyperparameter Tuning (AAAI 2020 · similarity 0.85)
- Batch Multi-Fidelity Bayesian Optimization with Deep Auto-Regressive Networks (NeurIPS 2021 · similarity 0.84)
- Efficient Hyperparameter Optimization with Adaptive Fidelity Identification (CVPR 2024 · similarity 0.83)
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.
