Research
Pruning Binarized Neural Networks: A Dedicated Framework and Globally Weighted Algorithms
arXiv:2608.26233v1 Announce Type: new Abstract: Extreme compression of deep neural networks, up to full binarization, dramatically reduces memory footprint and arithmetic complexity, facilitating depl
arXiv:2608.26233v1 Announce Type: new Abstract: Extreme compression of deep neural networks, up to full binarization, dramatically reduces memory footprint and arithmetic complexity, facilitating deployment on constrained edge hardware with field-programmable gate arrays (FPGAs) and microcontrollers. Although combining binarization with pruning promises additional efficiency gains, existing pruning strategies are ill-suited to binarized representations and rarely translate into meaningful hardware savings. We introduce a PyTorch-based, research-oriented framework that incorporates freezing and pruning mechanisms for designing and optimizing binarized neural networks. The framework enables rapid and reproducible evaluation of state-of-the-art approaches and the fast prototyping of new ones. Leveraging this framework, we propose a novel pruning method that accounts for the relative importance of learned parameters across abstraction levels. Such a global weighting mechanism consistently achieves a superior trade-off between model accuracy and pruning rate, achieving a 70% pruning rate on VGG11 with constant accuracy, while state-of-the-art results reach only 41% in the binarized setting.
Related
- Understanding Fault Tolerance of Adversarially Robust Pruned Models
- Investigating the Effect of Network Pruning on Performance and Interpretability
- Some Complexity Results for Robustness Verification for Binarized Neural Networks
- A Greedy Hierarchical Approach to Whole-Network Filter-Pruning in CNNs
Source: arXiv cs.LG | 2026-08-28