Research map / Architectures and efficiency / Neural architecture search
Multi-Objective Neural Architecture Search by Learning Search Space Partitions
Yiyang Zhao, Linnan Wang, Tian Guo
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 LaMOO space partitioning to qEHVI | qEHVI · NasBench201 | log hypervolume difference: 225% sample efficiency | Improved |
| add LaMOO space partitioning to CMA-ES | CMA-ES · NasBench201 | sample efficiency: 500% | Improved |
| add LaMOO to qEHVI | qEHVI · NasBench301 | hypervolume: 200% search performance improvement | Improved |
| add LaMOO to CMA-ES | CMA-ES · NasBench301 | sample efficiency: more than 250% | Improved |
| add LaMOO to CMA-ES | CMA-ES · HW-NAS-Bench (4 objectives) | log hypervolume difference | Improved |
| use few-shot instead of one-shot evaluation (ablation) | LaMOO · NasBench201 | log hypervolume difference | Improved |
| use vanilla training evaluation instead of few-shot (ablation) | LaMOO · NasBench201 | log hypervolume difference | Improved |
| use few-shot instead of one-shot LaMOONet (ablation) | LaMOONet · CIFAR-10 | test error (%): 2.85 -> 2.78 | Improved |
| use few-shot instead of one-shot LaMOONet (ablation) | LaMOONet · CIFAR-10 | GPU days: 1.18 -> 2.06 | Hurt |
| use LaMOO instead of vanilla NAS baselines (LaMOONet-P1) | NAS · CIFAR-10 | #parameters: 1.62M compared to at least 2.8M | Improved |
| use LaMOO instead of AmoebaNet search | NAS · CIFAR-10 | GPU days: 3150 -> 100 | Improved |
| use LaMOO one-shot instead of standard evaluation (ablation) | LaMOONet · CIFAR-10 | test error (%): 2.23 -> 2.85 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- DARTS — DARTS-: Robustly Stepping out of Performance Collapse Without Indicators (ICLR 2021) · better on 0 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Multi-objective Optimization by Learning Space Partition (ICLR 2022 · similarity 0.90)
- MIGO-NAS: Towards Fast and Generalizable Neural Architecture Search (TPAMI 2021 · similarity 0.89)
- Neural Architecture Generator Optimization (NeurIPS 2020 · similarity 0.88)
- Multi-objective Differentiable Neural Architecture Search (ICLR 2025 · similarity 0.88)
- Generalization Guarantees for Neural Architecture Search with Train-Validation Split (ICML 2021 · similarity 0.87)
- einspace: Searching for Neural Architectures from Fundamental Operations (NeurIPS 2024 · similarity 0.87)
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.
