Research map / Architectures and efficiency / Quantization
On the Impact of Weight Quantization on Deep Neural Network Uncertainty
Shuang Liang, Xun Lu, Zi-Ang Liu, Ming-Liang Wang, Yan Lyu, Shao-Qun Zhang
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 |
|---|---|---|---|
| apply MOMA to ternary MLP-2 training | T-MLP-2 · MNIST | uncertainty: 1.2572 -> 4.0129 | Improved |
| apply MOMA to ternary LeNet-5 training | T-LeNet-5 · MNIST | uncertainty: 2.8113 -> 4.0906 | Improved |
| apply MOMA to ternary VGG-13 training | T-VGG-13 · MNIST | uncertainty: 1.4500 -> 1.1463 | Improved |
| apply MOMA to ternary ResNet-18 training | T-ResNet-18 · MNIST | uncertainty: 4.2808 -> 2.4067 | Improved |
| apply MOMA to ternary VGG-13 training | T-VGG-13 · MNIST | accuracy: 0.9954 -> 0.9957 | Improved |
| apply MOMA to ternary ResNet-18 training | T-ResNet-18 · MNIST | accuracy: 0.9938 -> 0.9951 | Improved |
| apply MOMA to ternary MLP-2 training | T-MLP-2 · MNIST | accuracy: 0.9752 -> 0.9751 | No clear effect |
| apply MOMA to ternary LeNet-5 training | T-LeNet-5 · MNIST | accuracy: 0.9903 -> 0.9885 | Hurt |
| ternarize LeNet-5 weights | FP-LeNet-5 · MNIST | uncertainty: 1.7728 -> 2.8113 | Hurt |
| ternarize ResNet-18 weights | FP-ResNet-18 · MNIST | uncertainty: 6.9757 -> 4.2808 | Hurt |
| use EMP instead of MP for uncertainty estimation | MP · MNIST (FP-LeNet-5) | variance ratio vs MC: 1.1809 -> 1.0007 | Improved |
| use EMP instead of PL-DNN for uncertainty estimation | PL-DNN · CIFAR-10 (FP-LeNet-5) | variance ratio vs MC: 1.6167 -> 1.0098 | Improved |
Closest papers by meaning
- InfoQ: Mixed-Precision Quantization via Global Information Flow (AAAI 2026 · similarity 0.88)
- Explicit Loss-Error-Aware Quantization for Low-Bit Deep Neural Networks (CVPR 2018 · similarity 0.87)
- Towards Certificated Model Robustness Against Weight Perturbations (AAAI 2020 · similarity 0.87)
- Sparse Uncertainty Representation in Deep Learning with Inducing Weights (NeurIPS 2021 · similarity 0.87)
- Uncertainty Quantification with the Empirical Neural Tangent Kernel (NeurIPS 2025 · similarity 0.86)
- Epistemic Uncertainty Quantification For Pre-Trained Neural Networks (CVPR 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.
