Research map / Architectures and efficiency / Neural architecture search
KNAS: Green Neural Architecture Search
Jingjing Xu, Liang Zhao, Junyang Lin, Rundong Gao, Xu Sun, Hongxia Yang
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 |
|---|---|---|---|
| rank untrained candidate architectures using MGM and then train the top candidates | KNAS · ImageNet-16 | Acc: 45.05 vs 44.43 | Improved |
| rank untrained candidate architectures using MGM and then train the top candidates | KNAS · CIFAR10 | Acc: 93.43 vs 93.72 | Hurt |
| rank untrained candidate architectures using MGM and then train the top candidates | KNAS · CIFAR100 | Acc: 71.05 vs 72.12 | Hurt |
| rank untrained candidate architectures using MGM and then train the top candidates | KNAS · ImageNet-16 | Acc: 45.05 vs 45.01 | Improved |
| replace RoBERTA-large with the KNAS-searched architecture | text classification · MRPC | Acc: 92.08 -> 93.32 | Improved |
| replace RoBERTA-large with the KNAS-searched architecture | text classification · RTE | Acc: 83.51 -> 83.75 | Improved |
| increase the number of top MGM candidates from k=20 to k=40 (ablation) | KNAS · CIFAR10 | Acc: 93.38 -> 93.43 | Improved |
| increase the number of top MGM candidates from k=20 to k=40 (ablation) | KNAS · CIFAR100 | Acc: 70.78 -> 71.05 | Improved |
| increase the number of top MGM candidates from k=20 to k=40 (ablation) | KNAS · ImageNet-16 | Acc: 44.63 -> 45.05 | Improved |
| use MGM to rank randomly initialized architectures | KNAS · CIFAR100 | Spearman correlation with validation accuracy: 0.56 | Improved |
Closest papers by meaning
- Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective (ICLR 2021 · similarity 0.88)
- NAS-Bench-360: Benchmarking Neural Architecture Search on Diverse Tasks (NeurIPS 2022 · similarity 0.88)
- SGAS: Sequential Greedy Architecture Search (CVPR 2020 · similarity 0.88)
- DA-NAS: Data Adapted Pruning for Efficient Neural Architecture Search (ECCV 2020 · similarity 0.88)
- Data-Free Neural Architecture Search via Recursive Label Calibration (ECCV 2022 · similarity 0.88)
- Generic Neural Architecture Search via Regression (NeurIPS 2021 · similarity 0.88)
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.
