Research map / Architectures and efficiency / Pruning and efficient tuning
ChipNet: Budget-Aware Pruning with Heaviside Continuous Approximations
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 continuous Heaviside projection with crispness loss rather than logistic curves alone (ablation) | ChipNet channel pruning · CIFAR-100, WRN-26-12, 16x channel pruning | channel selection effectiveness | Helped |
| Train masks on Tiny ImageNet and transfer them to CIFAR-100 | ResNet-101, 40% channel budget · CIFAR-100 | accuracy: 71.6 -> 72.0 | Helped |
| Train masks on Tiny ImageNet and transfer them to CIFAR-100 | ResNet-101, 60% channel budget · CIFAR-100 | accuracy: 71.8 -> 72.1 | Helped |
| Train masks on Tiny ImageNet and transfer them to CIFAR-100 | ResNet-101, 20% channel budget · CIFAR-100 | accuracy: 71.3 -> 68.3 | Hurt |
| Perturb the tuned pruning hyperparameters by approximately 10% (ablation) | ChipNet, WRN-26-12 · CIFAR-100, 16x volume budget | accuracy: 0.7143; 0.7232; 0.72; 0.7132; 0.7168 | No clear effect |
| Prune using volume rather than channel constraint (ablation) | ChipNet, WRN-26-12 · WRN-26-12, matched remaining budgets | accuracy for the optimized constraint | Mixed |
Closest papers by meaning
- Structured Pruning of Neural Networks With Budget-Aware Regularization (CVPR 2019 · similarity 0.89)
- Pruning Neural Networks via Coresets and Convex Geometry: Towards No Assumptions (NeurIPS 2022 · similarity 0.89)
- DominoSearch: Find layer-wise fine-grained N:M sparse schemes from dense neural networks (NeurIPS 2021 · similarity 0.89)
- Pruning from Scratch (AAAI 2020 · similarity 0.88)
- Sparsest Models Elude Pruning: An Exposé of Pruning’s Current Capabilities (ICML 2024 · similarity 0.87)
- Lightweight Dataset Pruning without Full Training via Example Difficulty and Prediction Uncertainty (ICML 2025 · 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.