Research map / Architectures and efficiency / Neural architecture search
DATA: Differentiable ArchiTecture Approximation
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 |
|---|---|---|---|
| increase EGS sampling times M from 1 to 9 (ablation) | DATA · CIFAR-10 | test error (%): 2.94 -> 2.45 | Helped |
| increase sampling times M from 4 to 7 (ablation) | DATA · CIFAR-10 | test error (%): 2.70 -> 2.59 | Helped |
| increase sampling times M from 4 to 7 (ablation) | DATA · PTB | test perplexity: 56.2 -> 55.3 | Helped |
| increase sampling times M from 4 to 7 (ablation) | DATA · WikiText-2 | test perplexity: 64.6 -> 64.2 | Helped |
| increase sampling times M from 1 to 7 (ablation) | DATA · PASCAL VOC-2012 | mIOU (%): 73.4 -> 75.6 | Helped |
| increase sampling times M from 1 to 9 (ablation) | DATA · CIFAR-10 | parameters (M): 2.54 -> 3.97 | Hurt |
| use EGS multi-op sampling instead of one-hot selection (DARTS) to bridge search and child architecture gap | DARTS · CIFAR-10 validation error | search-child gap (%): 33.01 -> 1.87 | Helped |
| EGS vs Gumbel-Softmax one-hot (SNAS) | SNAS · CIFAR-10 validation error | search-child gap (%): 2.13 -> 1.87 | Helped |
| DATA M=7 vs DARTS second order on PTB | DARTS (second order) · PTB | test perplexity: 55.7 -> 55.3 | Helped |
| DATA M=4 vs DARTS second order on CIFAR-10 | DARTS (2-th order) · CIFAR-10 | test error (%): 2.76 -> 2.70 | No clear effect |
| DATA M=7 vs AmoebaNet-B on CIFAR-10 | AmoebaNet-B · CIFAR-10 | test error (%): 2.55 -> 2.59 | Hurt |
Closest papers by meaning
- BaLeNAS: Differentiable Architecture Search via the Bayesian Learning Rule (CVPR 2022 · similarity 0.89)
- DrNAS: Dirichlet Neural Architecture Search (ICLR 2021 · similarity 0.88)
- EG-NAS: Neural Architecture Search with Fast Evolutionary Exploration (AAAI 2024 · similarity 0.88)
- Rethinking Architecture Selection in Differentiable NAS (ICLR 2021 · similarity 0.88)
- FP-NAS: Fast Probabilistic Neural Architecture Search (CVPR 2021 · similarity 0.87)
- Insights from the Use of Previously Unseen Neural Architecture Search Datasets (CVPR 2024 · similarity 0.87)
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.