Research map / Architectures and efficiency / Quantization
Fast and Accurate Binary Neural Networks Based on Depth-Width Reshaping
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 |
|---|---|---|---|
| Reshape ResNet-18 and apply pruning | Binary ResNet-18 · CIFAR-10 | accuracy: 90.2 -> 91.3 | Helped |
| Use a DWR ResNet-20 backbone with ReActNet optimization | ReActNet · CIFAR-10 | accuracy: 87.5 -> 88.4 | Helped |
| Reshape VGG-11 and apply pruning | Binary VGG-11 · CIFAR-10 | accuracy: 83.6 -> 84.9 | Helped |
| Use a DWR ResNet-18 backbone with 2-step training | 2-step binary ResNet-18 · ImageNet | Top-1 accuracy: 65.5 -> 67.2 | Helped |
| Reshape ResNet-18 and apply pruning | Binary ResNet-18 · ImageNet | Top-1 accuracy: 55.5 -> 56.9 | Helped |
| Use a DWR ResNet-18 backbone with ReActNet optimization | ReActNet · ImageNet | Top-1 accuracy: 61.8 -> 62.5 | Helped |
| Add a convolutional layer to VGG-Small before reshaping and pruning (ablation) | Binary VGG-Small · CIFAR-10 | accuracy: 90.2 -> 90.6 | Helped |
| Remove the final convolutional layer from VGG-Small before reshaping and pruning (ablation) | Binary VGG-Small · CIFAR-10 | accuracy: 90.2 -> 90.1 | Hurt |
| Reshape ResNet-20 with s=3 without pruning (ablation) | Binary ResNet-20 · CIFAR-10 | accuracy: 85.2 -> 85.8 | Helped |
| Add pruning after ResNet-20 reshaping at s=3 (ablation) | Binary ResNet-20 · CIFAR-10 | accuracy: 85.8 -> 86.0 | Helped |
| Add one binary convolutional layer to ResNet-20 (ablation) | Binary ResNet-20 · CIFAR-10 | accuracy: 85.2 -> 85.2 | No clear effect |
| Increase pruning pre-training from 5 to 15 epochs (ablation) | DWR ResNet-20 + ReActNet · CIFAR-10 | accuracy: 88.0 -> 88.3 | Helped |
Closest papers by meaning
- BD-Net: Has Depth-Wise Convolution Ever Been Applied in Binary Neural Networks? (AAAI 2026 · similarity 0.90)
- Binarized Neural Networks (NeurIPS 2016 · similarity 0.88)
- A Main/Subsidiary Network Framework for Simplifying Binary Neural Networks (CVPR 2019 · similarity 0.88)
- Basic Binary Convolution Unit for Binarized Image Restoration Network (ICLR 2023 · similarity 0.87)
- High-Capacity Expert Binary Networks (ICLR 2021 · similarity 0.87)
- Training Binary Neural Networks via Gaussian Variational Inference and Low-Rank Semidefinite Programming (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.