Research map / Architectures and efficiency / RNNs and normalization
Rethinking Bottleneck Structure for Efficient Mobile Network Design
Daquan Zhou, Qibin Hou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan
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 sandglass block instead of bottleneck with depthwise conv in the bottleneck (ablation) | Bottleneck variant (a) · ImageNet | top-1 accuracy (%): 68.90 -> 74.02 | Improved |
| add a second 3x3 depthwise convolution in the bottleneck (ablation) | Bottleneck variant (a) · ImageNet | top-1 accuracy (%): 68.90 -> 70.11 | Improved |
| place depthwise convolutions in high-dimensional space with shortcuts between high-dimensional features (ablation) | Variant (b) · ImageNet | top-1 accuracy (%): 70.11 -> 74.02 | Improved |
| move depthwise conv to bottleneck positions of inverted residual block (ablation) | Inverted residual block · ImageNet | top-1 accuracy (%): 74.02 -> 69.26 | Hurt |
| use sandglass block instead of inverted residual block in MobileNetV2 | MobileNetV2 · ImageNet | top-1 accuracy (%): 72.3 -> 74.02 | Improved |
| add sandglass block to DARTS search space | DARTS · CIFAR-10 | test error (%): 3.11 -> 2.98 | Improved |
| add sandglass block to DARTS search space | DARTS · CIFAR-10 | parameters (M): 3.25 -> 2.45 | Improved |
| add inverted residual block to DARTS search space | DARTS · CIFAR-10 | test error (%): 3.11 -> 3.26 | Hurt |
Closest papers by meaning
- NetworkGym: Reinforcement Learning Environments for Multi-Access Traffic Management in Network Simulation (NeurIPS 2024 · similarity 0.78)
- TSNBench: Benchmarking LLM Proficiency in Time-Sensitive Networking (NeurIPS 2026 · similarity 0.74)
- Tackling the Data-Parallel Load Balancing Bottleneck in LLM Serving: Practical Online Routing at Scale (NeurIPS 2026 · similarity 0.74)
- Hierarchical Optimization via LLM-Guided Objective Evolution for Mobility-on-Demand Systems (NeurIPS 2025 · similarity 0.73)
- MAPS: Memory-Aware Predictive Scheduling Framework for Large Language Model Serving (ICML 2026 · similarity 0.73)
- CORE: Reducing UI Exposure in Mobile Agents via Collaboration Between Cloud and Local LLMs (NeurIPS 2025 · similarity 0.72)
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.
