Research map / Architectures and efficiency / Neural architecture search
Bridging the Gap between Sample-based and One-shot Neural Architecture Search with BONAS
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 a GCN architecture predictor instead of an MLP | BONAS predictor · NAS-Bench-101 test | correlation: 0.830 -> 0.841 | Helped |
| Use a GCN architecture predictor instead of a Meta NN | BONAS predictor · NAS-Bench-201 test | correlation: 0.967 -> 0.973 | Helped |
| Use a GCN architecture predictor instead of a Meta NN | BONAS predictor · LSTM-12K test | correlation: 0.582 -> 0.742 | Helped |
| Apply BONAS search in the NASNet space | CIFAR-10 NAS architecture search · CIFAR-10, NASNet search space | top-1 error (%): 2.64 -> 2.43 | Helped |
| Transfer BONAS-C's architecture rather than AmoebaNet-C's | ImageNet mobile architecture · ImageNet, mobile setting | top-1 error (%): 24.3 -> 24.6 | Hurt |
| Transfer BONAS-C's architecture rather than AmoebaNet-C's | ImageNet mobile architecture · ImageNet, mobile setting | top-5 error (%): 7.6 -> 7.5 | Helped |
| Replace EA candidate sampling with random sampling (ablation) | BONAS · NAS-Bench-201 | search efficiency | Hurt |
| Replace the GCN predictor with an LSTM predictor (ablation) | BONAS · NAS-Bench-201 | search efficiency | Hurt |
| Replace the GCN predictor with an MLP predictor (ablation) | BONAS · NAS-Bench-201 | search efficiency | Hurt |
| Add Bayesian sigmoid regression to the GCN selection score (ablation) | BONAS · NAS-Bench-201 | search efficiency | Helped |
| Use exponentially weighted loss instead of square loss (ablation) | BONAS · NAS-Bench-201 | search efficiency | Helped |
| Vary the GCN embedding size among 16, 32, 64, and 128 (ablation) | BONAS · NAS-Bench-201 | search efficiency | No clear effect |
Closest papers by meaning
- Distribution Consistent Neural Architecture Search (CVPR 2022 · similarity 0.92)
- Fitting the Search Space of Weight-sharing NAS with Graph Convolutional Networks (AAAI 2021 · similarity 0.91)
- BayesNAS: A Bayesian Approach for Neural Architecture Search (ICML 2019 · similarity 0.91)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.91)
- Posterior-Guided Neural Architecture Search (AAAI 2020 · similarity 0.91)
- NAS-Bench-1Shot1: Benchmarking and Dissecting One-shot Neural Architecture Search (ICLR 2020 · similarity 0.91)
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.