Research map / Architectures and efficiency / Neural architecture search
Adapting Neural Architectures Between Domains
Yanxi Li, Zhaohui Yang, Yunhe Wang, Chang Xu
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 hybrid (target) loss in both training and search phases (ablation) | AdaptNAS-S · ImageNet test | top-1 error (%): 25.3 -> 24.5 | Improved |
| add hybrid loss only in search phase (ablation) | AdaptNAS-S · ImageNet test | top-1 error (%): 25.3 -> 24.8 | Improved |
| use cosine instead of exponential gamma scheduler (ablation) | AdaptNAS-S · ImageNet test | top-1 error (%): 25.1 -> 25.3 | Hurt |
| use cosine instead of exponential gamma scheduler (ablation) | AdaptNAS-S · CIFAR-10 test | error (%): 2.93 -> 2.77 | Improved |
| mix source and target loss (alpha 0.75, beta 0.98) instead of source loss only (ablation) | AdaptNAS-C · ImageNet test | top-1 error (%): 25.7 -> 25.1 | Improved |
| target loss only (alpha 0) instead of hybrid at 1,000 target samples (beta 0.98) (ablation) | AdaptNAS-C · ImageNet test | top-1 error (%): 25.7 -> 29.5 | Hurt |
| add target samples via AdaptNAS-C (alpha 0.25, beta 0.50) vs source-only (alpha 1.00) (ablation) | AdaptNAS-C · ImageNet test | top-1 error (%): 25.4 -> 24.2 | Improved |
| add adversarial domain-distance loss during architecture search (AdaptNAS vs search on source) | DARTS-style NAS · MNIST to SVHN | target test accuracy (%): 94.70 -> 95.63 | Improved |
| add adversarial domain-distance loss (AdaptNAS vs search on source) | DARTS-style NAS · MNIST to MNIST-M | target test accuracy (%): 98.56 -> 98.75 | Improved |
| use Rot-4 instead of Rot-1 rotation task for AdaptNAS-C (ablation) | AdaptNAS-C · ImageNet test | top-1 error (%): 24.3 -> 24.2 | Improved |
| use Rot-4 instead of Rot-1 for AdaptNAS-S (ablation) | AdaptNAS-S · ImageNet test | top-1 error (%): 24.7 -> 25.3 | Hurt |
Closest papers by meaning
- Bridge the Gap Between Architecture Spaces via A Cross-Domain Predictor (NeurIPS 2022 · similarity 0.91)
- DA-NAS: Data Adapted Pruning for Efficient Neural Architecture Search (ECCV 2020 · similarity 0.91)
- NASTransfer: Analyzing Architecture Transferability in Large Scale Neural Architecture Search (AAAI 2021 · similarity 0.91)
- Neural Architecture Transfer (TPAMI 2021 · similarity 0.90)
- Insights from the Use of Previously Unseen Neural Architecture Search Datasets (CVPR 2024 · similarity 0.90)
- TransNAS-Bench-101: Improving Transferability and Generalizability of Cross-Task Neural Architecture Search (CVPR 2021 · 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.
