Research map / Architectures and efficiency / Neural architecture search
S2DNAS: Transforming Static CNN Model for Dynamic Inference via Neural Architecture Search
Zhihang Yuan, Bingzhe Wu, Guangyu Sun, Zheng Liang, Shiwan Zhao, Weichen Bi
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 |
|---|---|---|---|
| use S2DNAS instead of BranchyNet early exiting | ResNet-20 · CIFAR-10 | accuracy: from 91.37% to 91.41% | Improved |
| apply S2DNAS dynamic inference to the static model | ResNet-20 · CIFAR-10 | FLOPs: 41M -> 16M (61%) | Improved |
| apply S2DNAS dynamic inference to the static model | VGG16-BN · CIFAR-10 | FLOPs: 79% reduction with a slight accuracy drop | Improved |
| add BlockDrop controller for dynamic layer skipping | ResNet-20 · CIFAR-100 | FLOPs: 41M -> 53M (-29%) | Hurt |
| raise the exit thresholds (more computation) (ablation) | S2DNAS multi-stage models · CIFAR-10 and CIFAR-100 | accuracy | Improved |
| use MSDNet instead of S2DNAS at low budget | DenseNet-like model · CIFAR-10 | accuracy vs FLOPs: MSDNet better at FLOPs around 15M | Mixed |
Closest papers by meaning
- DS-Net++: Dynamic Weight Slicing for Efficient Inference in CNNs and Vision Transformers (TPAMI 2022 · similarity 0.86)
- Dynamic Neural Networks: A Survey (TPAMI 2021 · similarity 0.85)
- SNAS: stochastic neural architecture search (ICLR 2019 · similarity 0.85)
- Dynamic Convolutions Improve Transformers (NeurIPS 2026 · similarity 0.84)
- Revisiting Convolution Architecture in the Realm of DNA Foundation Models (ICLR 2025 · similarity 0.84)
- Inference-Scale Complexity in ANN-SNN Conversion for High-Performance and Low-Power Applications (CVPR 2025 · similarity 0.83)
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.
