Research map / Architectures and efficiency / Neural architecture search
Data-Free Neural Architecture Search via Recursive Label Calibration
Zechun Liu, Zhiqiang Shen, Yun Long, Eric Xing, Kwang-Ting Cheng, Chas Leichner
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 recursive label calibration and regional update to regularized synthesis (ablation) | synthetic data · SPOS search space, ImageNet | Spearman rank correlation: 0.87 -> 0.96 | Improved |
| synthesize data with regularization instead of random noise (ablation) | synthetic data · DARTS space, CIFAR-10 | accuracy rank correlation | Improved |
| add recursive label calibration to synthesized data (ablation) | synthetic data · DARTS space, CIFAR-10 | accuracy rank correlation | Improved |
| use label calibration when synthesizing data for DARTS search (ablation) | DARTS · CIFAR-10 | top-1 error: 3.29 -> 2.68 | Improved |
| use label calibration for SPOS search data (ablation) | SPOS · ImageNet | top-1 accuracy: 74.2 | Improved |
| replace original ImageNet with synthesized data for DARTS arch | DARTS · CIFAR-10 | top-1 error: 2.76 -> 2.68 | No clear effect |
| replace original data with synthesized data for SPOS | SPOS · ImageNet | top-1 error: 25.7 -> 25.8 | No clear effect |
| replace original data with synthesized data for ProxylessNAS | ProxylessNAS · ImageNet | top-1 error: 24.4 -> 24.2 | Improved |
| transfer data-free DARTS architecture to ImageNet | DARTS · ImageNet | top-1 error: 26.7 -> 26.4 | Improved |
| use our synthesis instead of ADI for data-free pruning | MetaPruning · ImageNet | top-1 error: 39.3 -> 36.5 | Improved |
| use our synthesis instead of ADI for knowledge transfer | student ResNet-18 · CIFAR-10 | top-1 error: 6.74 -> 5.97 | Improved |
| use our synthesis instead of ADI for knowledge transfer at 250 epochs | student ResNet-50 · ImageNet | top-1 error: 26.2 -> 25.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DARTS — DARTS-: Robustly Stepping out of Performance Collapse Without Indicators (ICLR 2021) · better on 2 of 2 shared results
- BlockQNN — BlockQNN: Efficient Block-Wise Neural Network Architecture Generation (TPAMI 2020) · better on 1 of 1 shared result
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
- NAS-Bench-101: Towards Reproducible Neural Architecture Search (ICML 2019 · similarity 0.92)
- Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective (ICLR 2021 · similarity 0.92)
- einspace: Searching for Neural Architectures from Fundamental Operations (NeurIPS 2024 · similarity 0.90)
- Insights from the Use of Previously Unseen Neural Architecture Search Datasets (CVPR 2024 · similarity 0.90)
- Towards Neural Architecture Search through Hierarchical Generative Modeling (ICML 2024 · similarity 0.90)
- Robustifying and Boosting Training-Free Neural Architecture Search (ICLR 2024 · similarity 0.90)
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.
