.png){kind=icon}
.png){kind=icon}
Return
Resource-Efficient and Layer Interdependence-Aware CNN Pruning Leveraging Filter Replacement
.png){kind=icon}
PRE
AI
S
M
M
M
K
DOI：10.1109/tnnls.2026.3658318.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Convolutional neural network (CNN) pruning has traditionally relied on heuristically designed importance criteria, often leading to limited generalizability and inconsistent performance. In this article, we propose a novel framework centered around <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">filter replacement</i> (FR), introducing pruning as a process of replacing selected filters with zero filters. Through a rigorous analysis, we derive an upper bound on the absolute error in the output of the subsequent layer and use this bound to define an efficient importance function. This importance function exhibits <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\gamma $ </tex-math></inline-formula>-weakly submodular properties, enabling the development of a simple, low-complexity, and data-free oblivious algorithm for selecting filters to prune. In addition, we extend the FR framework to include nonzero filter alternatives, leveraging a best-approximation technique to construct optimal replacements for the pruned filters. Extensive experiments on benchmark networks and datasets validate the effectiveness of our method. The proposed approach achieves state-of-the-art results, with a complexity comparable to basic techniques such as <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$l_{2}$ </tex-math></inline-formula>-norm pruning. Notably, our pruning method achieves 76.52% accuracy (ACC) in ResNet-50 on the ImageNet dataset, surpassing the baseline of 75.15%, while reducing network parameters by 25.5%. Our proposed resource efficiency (RE) metric assesses that the layer interdependence-aware pruning (LIAP) method is up to <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$10^{11}$ </tex-math></inline-formula> times more efficient than existing techniques, setting a new standard for resource-aware CNN pruning.
Keywords:
Best approximation
convolutional neural networks (CNNs)
filter pruning
model compression methods
Journal
IF:
8.9
Papers:
7.5K
Citations:
7.2W
Organization
Cited Papers