Research map / Architectures and efficiency / Pruning and efficient tuning
Interpretations Steered Network Pruning via Amortized Inferred Saliency Maps
Alireza Ganjdanesh, Shangqian Gao, Heng Huang
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 |
|---|---|---|---|
| add RBF-style mask parameterization instead of independent per-pixel masks (ablation) | ISP · CIFAR-10 ResNet-56 | test accuracy | Improved |
| use larger gamma_1 interpretation weight (instead of 0.1) (ablation) | ISP · CIFAR-10 ResNet-56 | test accuracy stability | Improved |
| use small gamma_2 FLOPs regularization weight (1.0) (ablation) | ISP · CIFAR-10 ResNet-56 | test accuracy | Mixed |
| add classification loss to interpretation-only objective (ablation) | ISP · CIFAR-10 ResNet-56 | test accuracy | Improved |
| add interpretation term to classification-loss-only objective (ablation) | ISP · CIFAR-10 ResNet-56 | test accuracy | Improved |
| replace EEMC with ISP | EEMC · CIFAR-10 ResNet-56 | delta accuracy: +0.06% -> +0.18% | Improved |
| replace DCP with ISP | DCP · CIFAR-10 MobileNetV2 | delta accuracy: +0.22% -> +0.32% | Improved |
| replace FPGM with ISP | FPGM · ImageNet ResNet-34 | delta Top-1: -1.29% -> -0.45% | Improved |
| replace EEMC with ISP | EEMC · ImageNet ResNet-50 | delta Top-1: -0.49% -> -0.16% | Improved |
| replace QI with ISP | QI · ImageNet ResNet-101 | delta Top-1: -2.31% -> +0.40% | Improved |
| replace MetaPruning with ISP | MetaPruning · ImageNet MobileNet-V2 | delta Top-1: -0.80% -> -0.15% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- HRank — HRank: Filter Pruning Using High-Rank Feature Map (CVPR 2020) · 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
- Rethinking the Smaller-Norm-Less-Informative Assumption in Channel Pruning of Convolution Layers (ICLR 2018 · similarity 0.89)
- Rethinking the Pruning Criteria for Convolutional Neural Network (NeurIPS 2021 · similarity 0.89)
- Pruning Convolutional Neural Networks for Resource Efficient Inference (ICLR 2017 · similarity 0.88)
- Efficient Distortion-Minimized Layerwise Pruning (TPAMI 2025 · similarity 0.88)
- Isomorphic Pruning for Vision Models (ECCV 2024 · similarity 0.88)
- UPDP: A Unified Progressive Depth Pruner for CNN and Vision Transformer (AAAI 2024 · similarity 0.87)
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.
