Research map / Architectures and efficiency / Quantization
MetaAug: Meta-Data Augmentation for Post-Training Quantization
Cuong Van Pham, Hoang Anh Dung, Cuong Cao Nguyen, Trung Le, Dinh Phung, Gustavo Carneiro, Thanh-Toan Do
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 |
|---|---|---|---|
| train transformation network with validation loss only (no preservation or margin loss) (ablation) | MetaAug · ImageNet W2A2 | top-1 accuracy (%): 53.71 -> 53.45 | Hurt |
| add distribution preservation loss L_DP instead of MSE (ablation) | MetaAug · ImageNet W2A2 | top-1 accuracy (%): 53.64 -> 54.09 | Improved |
| add margin loss to prevent identity mapping (ablation) | MetaAug with L_DP · ImageNet W2A2 | top-1 accuracy (%): 54.09 -> 54.22 | Improved |
| use MetaAug instead of Genie-M | Genie PTQ · ImageNet W2A2 | top-1 accuracy (%): 53.71 -> 54.22 | Improved |
| use learned transformation network instead of traditional augmentation (Mixup, Cutmix) (ablation) | Genie PTQ calibration data · ImageNet W2A2 ResNet-18 | top-1 accuracy (%): 54.15 -> 54.22 | Improved |
| apply photometric augmentation (brightness, contrast) to calibration data (ablation) | Genie-M · ImageNet W2A2 ResNet-18 | top-1 accuracy (%): 53.71 -> 53.35 | Hurt |
| combine MetaAug with Cutmix (ablation) | MetaAug · ImageNet W2A2 ResNet-18 | top-1 accuracy (%): 54.22 -> 54.63 | Improved |
| use MetaAug instead of QDrop in reducing train-test gap | PTQ · ImageNet W2A2 ResNet-18 | test accuracy (%): 51.14 -> 54.22 | Improved |
| use MetaAug* instead of Bit-Shrinking* | PTQ · ImageNet W4A4 ResNet-50 | top-1 accuracy (%): 76.04 -> 75.78 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Bit-Shrinking* — Bit-Shrinking: Limiting Instantaneous Sharpness for Improving Post-Training Quantization (CVPR 2023) · better on 5 of 6 shared results
- PD-Quant — PD-Quant: Post-Training Quantization Based on Prediction Difference Metric (CVPR 2023) · better on 1 of 1 shared result
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
- Learning to Quantize for Training Vector-Quantized Networks (ICML 2025 · similarity 0.88)
- Enhancing Post-training Quantization Calibration through Contrastive Learning (CVPR 2024 · similarity 0.88)
- ERQ: Error Reduction for Post-Training Quantization of Vision Transformers (ICML 2024 · similarity 0.86)
- AIQViT: Architecture-Informed Post-Training Quantization for Vision Transformers (AAAI 2025 · similarity 0.86)
- Learning from Loss Landscape: Generalizable Mixed-Precision Quantization via Adaptive Sharpness-Aware Gradient Aligning (ICML 2025 · similarity 0.86)
- Q-DiT: Accurate Post-Training Quantization for Diffusion Transformers (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.
