Research map / Architectures and efficiency / Neural architecture search
ZiCo: Zero-shot NAS via inverse Coefficient of Variation on Gradients
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 |
|---|---|---|---|
| Use ZiCo instead of #Params to rank architectures | zero-shot NAS on NATSBench-TSS CIFAR10 · NATSBench-TSS CIFAR10 | Kendall's τ correlation with test accuracy: 0.57 -> 0.61 | Helped |
| Use ZiCo instead of #Params to rank architectures | zero-shot NAS on NATSBench-TSS CIFAR100 · NATSBench-TSS CIFAR100 | Kendall's τ correlation with test accuracy: 0.55 -> 0.61 | Helped |
| Use ZiCo instead of #Params to rank architectures | zero-shot NAS on NATSBench-TSS Img16-120 · NATSBench-TSS Img16-120 | Kendall's τ correlation with test accuracy: 0.52 -> 0.60 | Helped |
| Search with ZiCo instead of #Params | ImageNet zero-shot NAS at 450M FLOPs · ImageNet, 450M FLOPs budget | Top-1 accuracy: 63.5 -> 78.1{$\pm$0.3} | Helped |
| Use two batches to compute ZiCo rather than varying the batch count (ablation) | ZiCo on NATSBench-TSS CIFAR100 · NATSBench-TSS CIFAR100 | Correlation with test accuracy | Mixed |
| Increase the ZiCo batch size until 64 rather than using smaller batches (ablation) | ZiCo on NATSBench-TSS CIFAR100 · NATSBench-TSS CIFAR100 | Correlation with test accuracy | Helped |
Closest papers by meaning
- NAS-Bench-Suite-Zero: Accelerating Research on Zero Cost Proxies (NeurIPS 2022 · similarity 0.92)
- L-SWAG: Layer-Sample Wise Activation with Gradients Information for Zero-Shot NAS on Vision Transformers (CVPR 2025 · similarity 0.91)
- ParZC: Parametric Zero-Cost Proxies for Efficient NAS (AAAI 2025 · similarity 0.91)
- Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions (AAAI 2025 · similarity 0.88)
- Revisiting Neural Networks for Few-Shot Learning: A Zero-Cost NAS Perspective (ICML 2025 · similarity 0.87)
- AZ-NAS: Assembling Zero-Cost Proxies for Network Architecture Search (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.