Research map / Architectures and efficiency / Neural architecture search
FBNet: Hardware-Aware Efficient ConvNet Design via Differentiable Neural Architecture Search
Bichen Wu, Xiaoliang Dai, Peizhao Zhang, Yanghan Wang, Fei Sun, Yiming Wu, Yuandong Tian, Peter Vajda, Yangqing Jia, Kurt Keutzer
Check 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 layer-wise differentiable search with target-device latency optimization instead of manual design | FBNet-A · ImageNet validation on Samsung Galaxy S8 | CPU latency: 21.7 ms -> 19.8 ms | Improved |
| Use layer-wise differentiable search with target-device latency optimization instead of manual design | FBNet-A · ImageNet validation | Top-1 accuracy (%): 72.0 -> 73.0 | Improved |
| Use layer-wise differentiable search with target-device latency optimization instead of RL-based search | FBNet-B · ImageNet validation | Top-1 accuracy (%): 74.0 -> 74.1 | Improved |
| Use layer-wise differentiable search with target-device latency optimization instead of RL-based search | FBNet-B · ImageNet validation on Samsung Galaxy S8 | CPU latency: 23.7 ms -> 23.1 ms | Improved |
| Search a model for the target input resolution and channel scaling rather than reuse MobileNetV2 structure | FBNet family · ImageNet with varied input resolution and channel scaling | Top-1 accuracy (%): 1.5% to 6.4% | Improved |
| Search a model for the target device rather than reuse a Samsung S8-optimized model | FBNet-iPhoneX · ImageNet on iPhone X | Latency: 27.53 ms -> 19.84 ms | Improved |
| Use a reduced-resolution, searched architecture | FBNet-96-0.35-1 · ImageNet (96, 0.35) | CPU latency: 3.8 ms -> 2.9 ms | Improved |
Later papers that compared with this method
1 later accepted paper reports FBNet-96-0.35-1 as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Butterfly Transform: An Efficient FFT Based Neural Architecture Design (CVPR 2020 · better on 2 of 2 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- FBNetV2: Differentiable Neural Architecture Search for Spatial and Channel Dimensions (CVPR 2020 · similarity 0.88)
- MnasNet: Platform-Aware Neural Architecture Search for Mobile (CVPR 2019 · similarity 0.87)
- MobileNetV4: Universal Models for the Mobile Ecosystem (ECCV 2024 · similarity 0.87)
- ElasticViT: Conflict-aware Supernet Training for Deploying Fast Vision Transformer on Diverse Mobile Devices (ICCV 2023 · similarity 0.87)
- Fast and Practical Neural Architecture Search (ICCV 2019 · similarity 0.85)
- 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.
