Research map / Architectures and efficiency / Neural architecture search
DPP-Net: Device-aware Progressive Search for Pareto-optimal Neural Architectures
Jin-Dong Dong, An-Chieh Cheng, Da-Cheng Juan, Wei Wei, Min Sun
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 |
|---|---|---|---|
| select models by accuracy only (PNAS criterion) instead of device-aware Pareto selection (ablation) | DPP-Net search · CIFAR-10 | mobile inference time (s): 0.912 vs 0.149 | Hurt |
| device-aware Pareto selection with workstation timing instead of PNAS criterion (ablation) | DPP-Net search · CIFAR-10 | error rate (%): 4.36 -> 4.78 | Hurt |
| add memory usage as fifth objective for mobile search (Panacea/M models) (ablation) | DPP-Net-M · CIFAR-10 | memory on mobile (Mem-M): 213MB -> 58MB | Improved |
| DPP-Net-Panacea instead of CondenseNet-86 | CIFAR-10 classifier · CIFAR-10 | error rate (%): 5.0 -> 4.62 ± 0.23 | Improved |
| DPP-Net-Panacea instead of CondenseNet-86 | CIFAR-10 classifier · CIFAR-10 | FLOPs: 65.8M -> 63.5M | Mixed |
| transfer DPP-Net-Panacea to ImageNet instead of CondenseNet-74 | ImageNet classifier · ImageNet | top-1 error (%): 26.2 -> 25.98 | Improved |
| DPP-Net-Panacea instead of NASNet (Mobile) | ImageNet classifier · ImageNet | top-1 error (%): 26.0 -> 25.98 | Improved |
| use DPP-Net-Panacea instead of CondenseNet-74 on mobile phone | ImageNet classifier · ImageNet | mobile inference time (s): 0.694 -> 0.676 | Improved |
| search on the target device itself rather than reusing another device's results | DPP-Net search · 4-layer cell models | inference time rank across devices | Improved |
| fewer parameters or FLOPs | DPP-Net models · CIFAR-10 | inference time | No clear effect |
| DPP-Net-ES (2.04M params) vs DPP-Net-WS (1.00M params) on embedded device (ablation) | DPP-Net · CIFAR-10 | embedded-system time (s): 0.075 -> 0.044 | Improved |
Closest papers by meaning
- MnasNet: Platform-Aware Neural Architecture Search for Mobile (CVPR 2019 · similarity 0.88)
- DDPNAS: Efficient Neural Architecture Search via Dynamic Distribution Pruning (IJCV 2023 · similarity 0.88)
- Distilling Optimal Neural Networks: Rapid Search in Diverse Spaces (ICCV 2021 · similarity 0.88)
- Multi-objective Differentiable Neural Architecture Search (ICLR 2025 · similarity 0.88)
- NPAS: A Compiler-Aware Framework of Unified Network Pruning and Architecture Search for Beyond Real-Time Mobile Acceleration (CVPR 2021 · similarity 0.87)
- MOTE-NAS: Multi-Objective Training-based Estimate for Efficient Neural Architecture Search (NeurIPS 2024 · similarity 0.85)
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.
