Research map / Architectures and efficiency / Neural architecture search
Progressive Differentiable Architecture Search: Bridging the Depth Gap Between Search and Evaluation
Xin Chen, Lingxi Xie, Jun Wu, Qi Tian
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 operation-level Dropout on skip-connect (M=2) (ablation) | P-DARTS search · CIFAR10 | test error (%): 2.93 -> 2.69 | Improved |
| add operation-level Dropout on skip-connect (M=4) (ablation) | P-DARTS search · CIFAR10 | test error (%): 3.51 -> 2.97 | Improved |
| add architecture refinement fixing skip-connects at 2 (ablation) | P-DARTS search · CIFAR10 | test error (%): 2.79 -> 2.65 | Improved |
| approximate search space from earlier stages instead of random sampling (ablation) | P-DARTS final stage · CIFAR10 | test error (%): 3.43 -> 2.58 | Improved |
| remove all skip-connects (M=0) versus M=1 (ablation) | P-DARTS cell · CIFAR10 | test error (%): 2.68 -> 2.78 | Hurt |
| increase kept skip-connects from 2 to 4 (ablation) | P-DARTS cell · CIFAR10 | test error (%): 2.69 -> 2.97 | Hurt |
| search at 17 cells in final stage instead of DARTS 8-cell search | P-DARTS · CIFAR10 | test error (%): 2.76 -> 2.50 | Improved |
| use P-DARTS instead of DARTS second order on ImageNet | P-DARTS · ImageNet | top-1 error (%): 26.7 -> 24.4 | Improved |
| search on CIFAR100 instead of CIFAR10 and evaluate on CIFAR100 | P-DARTS · CIFAR100 | test error (%): 16.55 -> 15.92 | Improved |
| search on CIFAR10 and evaluate on CIFAR10 instead of CIFAR100-searched | P-DARTS · CIFAR10 | test error (%): 2.62 -> 2.50 | Improved |
| initialize each stage from previous-stage weights (ablation) | P-DARTS · CIFAR10 | accuracy | Hurt |
| enlarge evaluation network channels from 36 to 64 | P-DARTS · CIFAR10 | test error (%): 2.50 -> 2.25 | Improved |
Closest papers by meaning
- Towards Improving the Consistency, Efficiency, and Flexibility of Differentiable Neural Architecture Search (CVPR 2021 · similarity 0.92)
- Cyclic Differentiable Architecture Search (TPAMI 2022 · similarity 0.90)
- DrNAS: Dirichlet Neural Architecture Search (ICLR 2021 · similarity 0.90)
- TF-NAS: Rethinking Three Search Freedoms of Latency-Constrained Differentiable Neural Architecture Search (ECCV 2020 · similarity 0.90)
- BaLeNAS: Differentiable Architecture Search via the Bayesian Learning Rule (CVPR 2022 · similarity 0.89)
- Understanding and Robustifying Differentiable Architecture Search (ICLR 2020 · 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.
