Research map / Architectures and efficiency / Neural architecture search
Learning Where to Look – Generative NAS Is Surprisingly Efficient
Jovita Lukasik, Steffen Jung, Margret Keuper
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 latent space optimization with weighted retraining (ablation) | AG-Net · NAS-Bench-101 | val accuracy: 94.38 -> 94.90 | Improved |
| add latent space optimization (ablation) | AG-Net · NAS-Bench-101 | test accuracy: 93.78 -> 94.18 | Improved |
| add latent space optimization (ablation) | AG-Net · NAS-Bench-201 CIFAR-100 | test accuracy: 71.83 -> 73.51 | Improved |
| add latent space optimization (ablation) | AG-Net · NAS-Bench-201 ImageNet16-120 | test accuracy: 45.04 -> 46.43 | Improved |
| add backpropagation through the surrogate into the generator (ablation) | AG-Net · NAS-Bench-101 | val accuracy: 94.71 -> 94.90 | Improved |
| add backpropagation through the surrogate into the generator (ablation) | AG-Net · NAS-Bench-201 CIFAR-100 | test accuracy: 73.22 -> 73.51 | Improved |
| add backpropagation through the surrogate (ablation) | AG-Net · NAS-Bench-201 CIFAR-10 | val accuracy: 91.60 -> 91.60 | No clear effect |
| replace XGB predictor with end-to-end learnable predictor (ablation) | AG-Net · NAS-Bench-101 | val accuracy: 94.61 -> 94.90 | Improved |
| add ranking loss to XGB predictor | AG-Net · NAS-Bench-101 | val accuracy: 94.61 -> 94.60 | No clear effect |
| replace XGB with AG-Net predictor | AG-Net · NAS-Bench-NLP | validation perplexity: 95.95 -> 95.86 | Hurt |
| increase query budget from 192 to 400 | AG-Net · NAS-Bench-201 CIFAR-10 | val accuracy: 91.60 -> 91.61 | Improved |
| increase queries from 192 to 400 | AG-Net · NAS-Bench-201 ImageNet16-120 | val accuracy: 46.64 -> 46.73 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- BANANAS — BANANAS: Bayesian Optimization with Neural Architectures for Neural Architecture Search (AAAI 2021) · better on 5 of 6 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
- Towards Neural Architecture Search through Hierarchical Generative Modeling (ICML 2024 · similarity 0.92)
- Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective (ICLR 2021 · similarity 0.91)
- einspace: Searching for Neural Architectures from Fundamental Operations (NeurIPS 2024 · similarity 0.91)
- Evolving Search Space for Neural Architecture Search (ICCV 2021 · similarity 0.91)
- Neural Architecture Generator Optimization (NeurIPS 2020 · similarity 0.91)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.89)
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.
