Deep Convolutional Neural Networks Structured Pruning via Gravity Regularization

Abdesselam Ferdi

REVIEW 4 major objections 5 minor 1 cited by

Deep Convolutional Neural Networks Structured Pruning via Gravity Regularization

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A gravity-inspired regularizer during training makes convolutional filters prunable at any ratio without retraining.

desk verdict The paper's headline no-retraining pruning claim is directly contradicted by its own Table I; the method is a minor variant of the author's earlier electrostatic force work, though the write-up is clear and the fine-tuned results are competitive. read the letter →

arxiv 2411.16901 v1 pith:CESHLE64 submitted 2024-11-25 cs.CV

Abdesselam Ferdi This is my paper · ORCID

classification cs.CV

keywords structuredpruningdeepconvolutionalneuralnetworksgravityregularizationL1normfilterCIFARResNet-56VGG-19

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that adding a gravity-inspired penalty to the training loss of a convolutional neural network prepares the network for structured pruning at any desired ratio, with no retraining and no change to the architecture. The penalty pulls filters far from the layer's largest $\ell^1$ filter toward zero weights while letting nearby filters keep their weights, so that after training a simple $\ell^1$-norm ranking identifies which filters to delete. If the claim holds, a single trained model could be deployed at several compression levels, and pruning stops being a separate post-training search-and-fine-tune stage. The paper reports experiments on ResNet-56 with CIFAR-10 and VGG-19 with CIFAR-100, comparing against existing structured pruning methods.

What carries the argument

The key object is the gravity regularizer $F_{n,l} = G m_{1,l} m_{n,l} / d_{n,l}^2$, appended to the standard loss as a penalty with strength $\alpha_g$. Filter mass is the $\ell^1$ norm of the filter weights, and the distance between filters is the reciprocal of their absolute index difference, so the regularizer is strongest for filters that the final pruning step will rank as least important. The attracting filter is the filter with the largest mass in the layer, which exerts no force on itself. This device converts structured pruning into a training-time weight redistribution: it uses the same $\ell^1$ norm both as 'mass' during training and as the pruning criterion afterward, so the ranking and the removal step are automatically aligned.

What would settle it

Train the same architecture twice with the same data and hyperparameters, but in the second run randomly permute the order of filters within each convolutional layer before gravity training begins; if pruning the two models at the same ratio yields essentially the same accuracy, the index-distance term is not responsible for the method's effect.

Watch

Extended reading notes

Core claim

The central discovery is that a Newton-style gravitational force, applied as a regularizer during training, can structure the weight distribution of convolutional filters so that filters located far from an attracting filter--the filter with the largest $\ell^1$ mass in the layer--are driven to zero, while filters close to it remain dense. The force is defined by $F = G m_1 m_n / d^2$, with mass $m$ equal to the $\ell^1$ norm of a filter's weights and with distance $d$ taken as the reciprocal of the absolute index difference between filters. Minimizing the regularized objective therefore multiplies a $\operatorname{sign}(w)$ penalty by $G m_1 / d^2$, making the penalty strongest for filters whose index lies far from the attracting filter. After training, the method prunes filters with the smallest $\ell^1$ norms at a uniform rate across layers; because the weight distribution already separates important from unimportant filters, the same trained model can be pruned at different ratios without retraining. The paper presents this as a way to configure a model for pruning during training rather than after it.

Load-bearing premise

The load-bearing premise is that a filter's position within its layer is a meaningful coordinate for pruning importance, so that filters far from the largest $\ell^1$ filter can be driven to zero without hurting accuracy.

Editorial extensions

If this is right

A gravity-trained model can be compressed to different sizes from a single training run, since the pruning ratio is chosen after training and no retraining is needed when the ratio changes.
The method applies to both randomly initialized and pretrained weights, and requires no modification to the original architecture.
The pruning step is a local $\ell^1$-norm ranking within each layer, so implementation is simple and the reported speedup and compression ratios grow with the pruning ratio.
The training overhead is nontrivial, about 80 percent longer than baseline for ResNet-56 on CIFAR-10 in the reported settings, so the benefit is a tradeoff against training cost.
After optional fine-tuning, the pruned models are competitive with several existing structured pruning methods at comparable speedups on CIFAR-10 and CIFAR-100.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If the method works as described, the index-distance geometry may be replaceable: any ordering that assigns larger penalties to filters the network can afford to lose would produce the same effect, so the physics analogy could be swapped for a simpler per-filter penalty schedule.
The 'no retraining' claim is demonstrated only for small CIFAR models; extrapolating to ImageNet-scale networks or to detection and segmentation heads would require testing, since redundancy patterns differ.
A direct test of the distance mechanism would be to randomly permute filter indices before gravity training; if pruning accuracy is unchanged, the attraction is not coming from index distance.
The method's value over standard $\ell^1$ regularization may come from its position-dependent penalty rates: filters in different index neighborhoods are penalized differently, which can be seen as a hand-designed sparsity curriculum.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The key thing to know: the paper's central promise — prune at any ratio without retraining while preserving accuracy — is not just unproven; it is refuted by the paper's own Table I. Pruning ResNet-56/CIFAR-10 at only 10% drops accuracy from 93.39% to 74.52%, and VGG-19/CIFAR-100 drops from 73.76% to 63.39% at the same ratio. Those are not "preserving accuracy" numbers, and the Discussion even concedes that 40% pruning costs roughly 37% accuracy. So the core contribution, as stated, collapses on the evidence provided.

What the paper does well: it is clearly written, gives precise pseudocode for training and pruning, and is upfront about training overhead (80% slower for ResNet-56). The fine-tuned results in Tables IV and V are competitive with existing methods, and the authors include a hyperparameter sweep. The mathematical description is internally consistent once you notice that d in Eq. 4 is the reciprocal of the index distance, so the "larger d" language in Section VI.C is not an inconsistency — the stress-test note claiming otherwise misreads that.

The soft spots are substantial. The no-retraining claim is the reason to care, and it fails. The fine-tuned comparisons use weakened baselines: the gravity-trained VGG-19 baseline is 70.99% versus roughly 73.8% for other methods, so the accuracy drop in Table IV is measured from a lower starting point, flattering the result. The gravity-rate hyperparameter is selected by inspecting Figure 2, which is the very curve used to report the headline pruned accuracies — that is fitting to the evaluation. No code, seeds, or error bars are supplied, so Figure 2 cannot be checked. Novelty is thin: this is essentially the author's earlier electrostatic-force regularization (Ref. [2]) with Coulomb's law swapped for Newton's, which is the same inverse-square structure on index distance. The index distance itself is asserted as a meaningful coordinate without any supporting analysis.

Who is this for? Someone cataloging simple training-time regularizers for filter pruning might find the idea worth a footnote, but they should not rely on the no-retraining claim. The paper is a coherent write-up of a plausible-sounding method that its own experiments show does not work as advertised. A serious editor would desk-reject it: the primary contribution is self-contradicted, and the residual fine-tuned results are not novel enough to justify referee time.

Referee Report

4 major / 5 minor

Summary. This manuscript proposes a structured pruning method for DCNNs based on a 'gravity' regularizer added during training. Filters are assigned masses equal to their L1 norms, an attracting filter is chosen as the largest-mass filter, and a penalty term proportional to the product of masses and inversely proportional to a squared distance (defined as the reciprocal of the index difference) is added to the loss. The authors claim that after training, filters can be pruned at arbitrary ratios without retraining. They report experiments on CIFAR-10/100 with ResNet-56 and VGG-19, including pruning-only results (Table I), fine-tuned comparisons with existing methods (Tables III–IV), and ablation studies over the gravity rate. The paper concludes that the method achieves competitive results while eliminating the fine-tuning stage.

Significance. If the no-retraining claim were true, this would be an attractive, architecture-agnostic approach because it would decouple the pruning ratio from the training procedure. However, the paper provides no code, no error bars, and no seed information; its only evidence for the headline claim is a set of curves (Figure 2) that cannot be checked. The fine-tuned comparisons in Tables III and IV are weakened by the fact that gravity training itself lowers the baseline accuracy, so the reported 'accuracy drop' values are relative to a weaker model rather than to the standard baselines used by the comparison methods. The manuscript therefore does not currently provide a sound basis for its claimed advantage.

major comments (4)

[V.A, Table I] The central claim that the method preserves accuracy without retraining is contradicted by the paper's own Table I: pruning ResNet-56/CIFAR-10 at 10% drops accuracy from 93.39% to 74.52% (an 18.87-point drop), and VGG-19/CIFAR-100 drops from 73.76% to 63.39% (a 10.37-point drop). Section VI.A concedes that 40% pruning reduces baseline accuracy by about 37%. The only support for the no-retraining claim is Figure 2, which is presented without error bars, seeds, or code; the headline claim must be either removed or replaced by evidence from a reproducible experimental protocol.
[V.C, Tables III and IV] The comparisons against existing methods are made after fine-tuning, which is exactly the stage the method claims to eliminate, and they compare accuracy drops relative to weakened gravity-trained baselines. For example, the VGG-19 Gravity (p) baseline is 70.99%, 2.77 points below the standard VGG-19 baseline of 73.76% used in Table I, and the ResNet-56 Gravity (r) baseline is 88.02% versus roughly 93.5%. Several 'accuracy drops' are therefore negative (e.g., Gravity (r) at 2.17x speedup in Table III), which does not indicate improved pruning but rather recovery from a degraded baseline after fine-tuning. These comparisons do not support the claim of 'minimal accuracy drop compared to all existing methods.'
[VI.D and Figure 2] The gravity rate α_g is selected as the value yielding the best pruning curves in Figure 2 (Section VI.D), and the same curves are used to demonstrate the method's headline no-retraining performance. No separate validation set or explicit model-selection procedure is described, so the reported pruning results are optimistically biased and the sensitivity of the method to α_g is not honestly assessed.
[III.B and VI.C] The distance formulation is internally inconsistent. Equation (4) defines d = 1/|p1-pn|, making the penalty in Eq. (5) proportional to |p1-pn|²; yet Section VI.C states that 'When the distance d is larger, the penalty term decreases' and 'when d is smaller, the penalty term increases.' Under Eq. (4), a larger d corresponds to a smaller index separation, so the explanation is at odds with the usual meaning of distance and the derivation of the pruning behavior is not self-consistent. In addition, using the filter index as a spatial coordinate is not justified because filters in a convolutional layer are permutation-invariant; an index-based distance defines an arbitrary ordering with no known relation to functional importance.

minor comments (5)

[III.B, Eq. (9)] The gradient derivation treats m1 and d as constants when differentiating the penalty term, but the attracting filter (and hence m1 and the index ordering) can change during training; the authors should state explicitly that this is an approximation or discuss its effect on the learning dynamics.
[IV.A] The reference to the CIFAR dataset cites reference [2] (a paper on electrostatic-force pruning) rather than a dataset reference; the dataset citation is missing.
[IV.B] Section IV.B says the fine-tuning uses 'the same hyperparameters as other pruning methods [7]' but does not specify what those hyperparameters are; a concrete list is needed for reproducibility.
[V.A, Figure 1 and 2] Figure 1 is not referenced in the text, and Figure 2 is never described in terms of the number of runs or variance; both should be integrated into the narrative with proper captions.
[III.B] The notation d is used both for the index difference in Eq. (3) and its reciprocal in Eq. (4); distinct symbols or an explicit naming convention would remove avoidable ambiguity.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the gravity regularizer is a self-contained weighted-L1 penalty, and the reported pruning accuracy is an empirical outcome rather than an identity.

full rationale

The derivation chain is self-contained. Section III.B defines filter mass as the L1 norm (Eq. 2), index distance (Eq. 3), reformulates it as its reciprocal (Eq. 4), and inserts it into a gravity-style force (Eq. 5) that is added to the training loss (Eq. 6). The gradient update in Eq. 10 is exactly the derivative of that penalty: a per-filter L1-norm gradient scaled by alpha_g G m1 / d^2. Nothing in the derivation assumes the headline conclusion that pruning preserves accuracy; whether removing low-L1 filters after such training is safe is an empirical question settled by Tables I-IV and Figure 2. The only mild methodological concern is that alpha_g is chosen by inspecting Figure 2, the same no-fine-tuning pruning curves used as evidence, and the physics terms are a relabeling of a distance-weighted L1 penalty; these are presentation and evaluation-practice issues, not circular reduction of the claimed result to its inputs. Self-citations to the author's prior electrostatic-force work [2] are comparative and not load-bearing. Empirical weaknesses, such as the large accuracy drops in Table I, are correctness concerns rather than circularity.

Assumptions & free parameters 2 free parameters · 3 assumptions · 1 invented entities

The method depends on a hyperparameter (alpha_g) fitted to the reported evaluation, a domain assumption about index distance, and a non-differentiable choice of the attracting filter. No independent evidence supports the physics analogy.

free parameters (2)

gravity rate alpha_g = 1e5 for ResNet-56/CIFAR-10, 1e2 for VGG-19/CIFAR-100
Selected by inspecting the pruning accuracy curves in Figure 2, which are the same results reported in the paper; this is tuning on the evaluation data.
gravitational constant G = 6.7e-11 (physical constant)
The standard gravitational constant is used, but in practice it is absorbed into the product alpha_g * G; the effective regularization strength is a single fitted hyperparameter.

assumptions (3)

domain assumption Filter index within a layer provides a meaningful geometric prior: filters far apart in index should be pruned, filters close should be kept.
Introduced in Section III.B, where distance is defined as the absolute difference of filter indices. No empirical or theoretical support that filter position in the weight tensor correlates with pruning importance.
domain assumption The L1 norm of a filter measures its importance and is a stable attractor (the largest-L1 filter can serve as the attracting filter).
The mass is defined as the L1 norm (Eq. 2) and the attracting filter is the largest-L1 filter (Algorithm 1). This is a common heuristic in pruning, but the paper provides no analysis of its stability during training.
ad hoc to paper The gradient of the L1 norm can be treated as the sign function and the attracting filter's identity can be treated as fixed for backpropagation.
Eq. 8 to Eq. 9 implicitly ignores the derivative of m1 with respect to the weights and the argmax selection of the attracting filter, which is non-differentiable.

invented entities (1)

Gravitational force between convolution filters
purpose: A penalty term that drives filter weights toward zero based on distance from the largest-L1 filter, enabling structured pruning.
This 'force' is a metaphor for a distance-weighted L1 regularizer; it has no falsifiable handle outside the method itself.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Deep Convolutional Neural Networks Structured Pruning via Gravity Regularization." pith.science (2026). https://pith.science/paper/CESHLE64

@misc{pith2026241116901,
  author       = {Pith},
  title        = {Pith review of: Deep Convolutional Neural Networks Structured Pruning via Gravity Regularization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CESHLE64}},
  note         = {Machine review of arXiv:2411.16901}
}

read the original abstract

Structured pruning is a widely employed strategy for accelerating deep convolutional neural networks (DCNNs). However, existing methods often necessitate modifications to the original architectures, involve complex implementations, and require lengthy fine-tuning stages. To address these challenges, we propose a novel physics-inspired approach that integrates the concept of gravity into the training stage of DCNNs. In this approach, the gravity is directly proportional to the product of the masses of the convolution filter and the attracting filter, and inversely proportional to the square of the distance between them. We applied this force to the convolution filters, either drawing filters closer to the attracting filter (experiencing weaker gravity) toward non-zero weights or pulling filters farther away (subject to stronger gravity) toward zero weights. As a result, filters experiencing stronger gravity have their weights reduced to zero, enabling their removal, while filters under weaker gravity retain significant weights and preserve important information. Our method simultaneously optimizes the filter weights and ranks their importance, eliminating the need for complex implementations or extensive fine-tuning. We validated the proposed approach on popular DCNN architectures using the CIFAR dataset, achieving competitive results compared to existing methods.

Figures

Figures reproduced from arXiv: 2411.16901 by the authors.

**Figure 2.** In this figure, for each network, the pruned top-1 [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 2.** Top-1 accuracy of pruned ResNet-56 and VGG-19 models, initialized with pretrained weights and trained with gravity [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Lightweight G-YOLOv11: Advancing Efficient Fracture Detection in Pediatric Wrist X-rays
eess.IV 2024-12 conditional novelty 3.0 of 10

G-YOLOv11l, a ghost-convolution and halved-filter YOLOv11 variant, reaches mAP@0.5 of 0.535 at 2.4 ms per image, 68.7% smaller than YOLOv11l but 13.6% less accurate.

Reference graph

Works this paper leans on

19 extracted references · 16 canonical work pages · cited by 1 Pith paper

[2]

Electrostatic force regularization for neural structured pruning,

Abdesselam Ferdi, Abdelmalik Taleb-Ahmed, Amir Nakib, and Youcef Ferdi, “Electrostatic force regularization for neural structured pruning,” arXiv preprint arXiv:2411.11079 , 2024

arXiv 2024
[1]

Complexity- driven model compression for resource-constrained deep learning on edge,

Muhammad Zawish, Steven Davy, and Lizy Abraham, “Complexity- driven model compression for resource-constrained deep learning on edge,” IEEE Transactions on Artificial Intelligence , 2024

work page 2024
[3]

Torque based structured pruning for deep neural network,

Arshita Gupta, Tien Bau, Joonsoo Kim, Zhe Zhu, Sumit Jha, and Hrishikesh Garud, “Torque based structured pruning for deep neural network,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2024, pp. 2711–2720

work page 2024
[4]

Accelerating deep neural networks via semi-structured activation sparsity,

Matteo Grimaldi, Darshan C. Ganji, Ivan Lazarevich, and Sudhakar Sah, “Accelerating deep neural networks via semi-structured activation sparsity,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops , October 2023, pp. 1179–1188

work page 2023
[5]

Advancing model pruning via bi-level optimization,

Yihua Zhang, Yuguang Yao, Parikshit Ram, Pu Zhao, Tianlong Chen, Mingyi Hong, Yanzhi Wang, and Sijia Liu, “Advancing model pruning via bi-level optimization,” Advances in Neural Information Processing Systems, vol. 35, pp. 18309–18326, 2022

work page 2022
[6]

Google colaboratory: Online jupyter notebooks,

Google Colaboratory, “Google colaboratory: Online jupyter notebooks,” 2024, Accessed: 2024-10-1

work page 2024
[7]

Neural pruning via growing regularization,

Huan Wang, Can Qin, Yulun Zhang, and Yun Fu, “Neural pruning via growing regularization,” in International Conference on Learning Representations (ICLR), 2021

work page 2021
[8]

Channel pruning for accelerating very deep neural networks,

Yihui He, Xiangyu Zhang, and Jian Sun, “Channel pruning for accelerating very deep neural networks,” in Proceedings of the IEEE international conference on computer vision , 2017, pp. 1389–1397

work page 2017

Show all 19 references

[9]

Amc: Automl for model compression and acceleration on mobile devices,

Yihui He, Ji Lin, Zhijian Liu, Hanrui Wang, Li-Jia Li, and Song Han, “Amc: Automl for model compression and acceleration on mobile devices,” in Proceedings of the European conference on computer vision (ECCV), 2018, pp. 784–800

2018
[10]

Soft filter pruning for accelerating deep convolutional neural networks,

Yang He, Guoliang Kang, Xuanyi Dong, Yanwei Fu, and Yi Yang, “Soft filter pruning for accelerating deep convolutional neural networks,”arXiv preprint arXiv:1808.06866, 2018

2018 arXiv
[11]

Whc: Weighted hybrid criterion for filter pruning on convolutional neural networks,

Shaowu Chen, Weize Sun, and Lei Huang, “Whc: Weighted hybrid criterion for filter pruning on convolutional neural networks,” in ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE, 2023, pp. 1–5

2023
[12]

Learning filter pruning criteria for deep convolutional neural networks acceleration,

Yang He, Yuhang Ding, Ping Liu, Linchao Zhu, Hanwang Zhang, and Yi Yang, “Learning filter pruning criteria for deep convolutional neural networks acceleration,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 2009–2018

2020
[13]

Channel pruning via automatic structure search,

Mingbao Lin, Rongrong Ji, Yuxin Zhang, Baochang Zhang, Yongjian Wu, and Yonghong Tian, “Channel pruning via automatic structure search,” arXiv preprint arXiv:2001.08565 , 2020

2001 arXiv
[14]

Channel pruning via lookahead search guided reinforcement learning,

Zi Wang and Chengcheng Li, “Channel pruning via lookahead search guided reinforcement learning,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2022, pp. 2029–2040

2022
[15]

Cen- tripetal sgd for pruning very deep convolutional networks with com- plicated structure,

Xiaohan Ding, Guiguang Ding, Yuchen Guo, and Jungong Han, “Cen- tripetal sgd for pruning very deep convolutional networks with com- plicated structure,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2019, pp. 4943–4953

2019
[16]

Auto- balanced filter pruning for efficient convolutional neural networks,

Xiaohan Ding, Guiguang Ding, Jungong Han, and Sheng Tang, “Auto- balanced filter pruning for efficient convolutional neural networks,” in Proceedings of the AAAI Conference on Artificial Intelligence , 2018, vol. 32

2018
[17]

Eigen- damage: Structured pruning in the kronecker-factored eigenbasis,

Chaoqi Wang, Roger Grosse, Sanja Fidler, and Guodong Zhang, “Eigen- damage: Structured pruning in the kronecker-factored eigenbasis,” in International conference on machine learning . PMLR, 2019, pp. 6566– 6575

2019
[18]

Quadratic convolution-based yolov8 (q- yolov8) for localization of intracranial hemorrhage from head ct images,

Abdesselam Ferdi, Said Benierbah, Amir Nakib, Youcef Ferdi, and Abdelmalik Taleb-Ahmed, “Quadratic convolution-based yolov8 (q- yolov8) for localization of intracranial hemorrhage from head ct images,” Biomedical Signal Processing and Control , vol. 96, pp. 106611, 2024

2024
[19]

Residual encoder- decoder based architecture for medical image denoising,

Abdesselam Ferdi, Said Benierbah, and Amir Nakib, “Residual encoder- decoder based architecture for medical image denoising,” Multimedia Tools and Applications, pp. 1–18, 2024

2024

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [2]

Electrostatic force regularization for neural structured pruning,

Abdesselam Ferdi, Abdelmalik Taleb-Ahmed, Amir Nakib, and Youcef Ferdi, “Electrostatic force regularization for neural structured pruning,” arXiv preprint arXiv:2411.11079 , 2024

arXiv 2024

[2] [1]

Complexity- driven model compression for resource-constrained deep learning on edge,

Muhammad Zawish, Steven Davy, and Lizy Abraham, “Complexity- driven model compression for resource-constrained deep learning on edge,” IEEE Transactions on Artificial Intelligence , 2024

work page 2024

[3] [3]

Torque based structured pruning for deep neural network,

Arshita Gupta, Tien Bau, Joonsoo Kim, Zhe Zhu, Sumit Jha, and Hrishikesh Garud, “Torque based structured pruning for deep neural network,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2024, pp. 2711–2720

work page 2024

[4] [4]

Accelerating deep neural networks via semi-structured activation sparsity,

Matteo Grimaldi, Darshan C. Ganji, Ivan Lazarevich, and Sudhakar Sah, “Accelerating deep neural networks via semi-structured activation sparsity,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops , October 2023, pp. 1179–1188

work page 2023

[5] [5]

Advancing model pruning via bi-level optimization,

Yihua Zhang, Yuguang Yao, Parikshit Ram, Pu Zhao, Tianlong Chen, Mingyi Hong, Yanzhi Wang, and Sijia Liu, “Advancing model pruning via bi-level optimization,” Advances in Neural Information Processing Systems, vol. 35, pp. 18309–18326, 2022

work page 2022

[6] [6]

Google colaboratory: Online jupyter notebooks,

Google Colaboratory, “Google colaboratory: Online jupyter notebooks,” 2024, Accessed: 2024-10-1

work page 2024

[7] [7]

Neural pruning via growing regularization,

Huan Wang, Can Qin, Yulun Zhang, and Yun Fu, “Neural pruning via growing regularization,” in International Conference on Learning Representations (ICLR), 2021

work page 2021

[8] [8]

Channel pruning for accelerating very deep neural networks,

Yihui He, Xiangyu Zhang, and Jian Sun, “Channel pruning for accelerating very deep neural networks,” in Proceedings of the IEEE international conference on computer vision , 2017, pp. 1389–1397

work page 2017