Research map / Architectures and efficiency / RNNs and normalization
Resilient Binary Neural Network
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 |
|---|---|---|---|
| Use ReBNN instead of ReCU | ResNet-18, two-stage training · ImageNet | Top-1 accuracy: 66.4 -> 66.9 | Helped |
| Use ReBNN instead of ReActNet | ResNet-34, two-stage training · ImageNet | Top-1 accuracy: 69.3 -> 69.9 | Helped |
| Use ReBNN instead of BiDet | Faster-RCNN with ResNet-18 · COCO minival | mAP@[.5,.95]: 15.7 -> 19.6 | Helped |
| Use ReBNN instead of BiDet | SSD with VGG-16 · COCO minival | mAP@[.5,.95]: 13.2 -> 18.1 | Helped |
| Use ReBNN instead of BiBERT | BERT_BASE · GLUE development set | Avg.: 63.2 -> 65.8 | Helped |
| Set constant γ to 1e-5 rather than 0 (ablation) | ReBNN · ImageNet, ResNet-18 gamma ablation | Top-1 accuracy: 65.8 -> 66.2 | Helped |
| Set constant γ to 1e-4 rather than 0 (ablation) | ReBNN · ImageNet, ResNet-18 gamma ablation | Top-1 accuracy: 65.8 -> 66.4 | Helped |
| Set constant γ to 1e-3 rather than 0 (ablation) | ReBNN · ImageNet, ResNet-18 gamma ablation | Top-1 accuracy: 65.8 -> 66.3 | Helped |
| Set constant γ to 1e-2 rather than 0 (ablation) | ReBNN · ImageNet, ResNet-18 gamma ablation | Top-1 accuracy: 65.8 -> 65.9 | No clear effect |
| Calculate γ from the maximum intra-channel gradient only (ablation) | ReBNN · ImageNet, ResNet-18 gamma ablation | Top-1 accuracy: 65.8 -> 66.3 | Helped |
| Calculate γ using the sign-change proportion and maximum gradient (ablation) | ReBNN · ImageNet, ResNet-18 gamma ablation | Top-1 accuracy: 66.3 -> 66.9 | Helped |
Closest papers by meaning
- SA-BNN: State-Aware Binary Neural Network (AAAI 2021 · similarity 0.89)
- How Do Adam and Training Strategies Help BNNs Optimization (ICML 2021 · similarity 0.88)
- Latent Weights Do Not Exist: Rethinking Binarized Neural Network Optimization (NeurIPS 2019 · similarity 0.87)
- Fast and Accurate Binary Neural Networks Based on Depth-Width Reshaping (AAAI 2023 · similarity 0.87)
- Training Binary Neural Networks via Gaussian Variational Inference and Low-Rank Semidefinite Programming (NeurIPS 2024 · similarity 0.86)
- Can DBNNs Robust to Environmental Noise for Resource-constrained Scenarios? (ICML 2025 · similarity 0.86)
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.