Research map / Architectures and efficiency / Quantization
Instance-Aware Dynamic Neural Network Quantization
Check 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 per-image dynamic mixed precision instead of static DoReFa quantization | ResNet-20 · CIFAR-10 | Top-1 accuracy (%): 89.9 -> 90.23; 90.5 -> 90.79; 90.4 -> 91.09 | Helped |
| Use per-image dynamic mixed precision instead of static PACT quantization | ResNet-20 · CIFAR-10 | Top-1 accuracy (%): 91.1 -> 91.38; 91.3 -> 91.60; 91.7 -> 92.01 | Mixed |
| Use per-image dynamic mixed precision instead of static DoReFa quantization | ResNet-50 · ImageNet | Top-1 accuracy (%): 69.9 -> 71.81 | Helped |
| Use per-image dynamic mixed precision instead of static PACT quantization | ResNet-50 · ImageNet | Top-1 accuracy (%): 76.5 -> 76.94 | Helped |
| Use per-image dynamic mixed precision instead of static LQ-Nets quantization | ResNet-50 · ImageNet | Top-1 accuracy (%): 74.2 -> 74.89 | Helped |
| Use per-image dynamic mixed precision instead of static LSQ quantization | ResNet-50 · ImageNet | Top-1 accuracy (%): 76.7 -> 77.02 | Helped |
| Use a shared early feature extractor rather than separate bit-controller convolution layers (ablation) | DQNet · ImageNet (ResNet-18 ablation) | Top-1 accuracy (%) | No clear effect |
| Use only one convolutional layer in the bit-controller (ablation) | DQNet · ImageNet (ResNet-18 ablation) | Top-1 accuracy (%) | Hurt |
| Increase the number of shared layers in the bit-controller (ablation) | DQNet · ImageNet (ResNet-18 ablation) | Top-1 accuracy (%) | Mixed |
Closest papers by meaning
- DQT: Dynamic Quantization Training via Dequantization-Free Nested Integer Arithmetic (AAAI 2026 · similarity 0.89)
- Adaptive Quantization for Deep Neural Network (AAAI 2018 · similarity 0.88)
- Quantization Networks (CVPR 2019 · similarity 0.88)
- Efficient Quantization for Neural Networks with Binary Weights and Low Bitwidth Activations (AAAI 2019 · similarity 0.87)
- Distribution-Aware Adaptive Multi-Bit Quantization (CVPR 2021 · similarity 0.87)
- Quantization without Tears (CVPR 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.