This work characterizes the fundamental limit of network pruning using statistical dimension and convex geometry.
problem The fundamental limit of network pruning is still lacking, especially for deep neural networks.
method Directly imposing sparsity constraint on the loss function and using statistical dimension in convex geometry.
result Characterizes the sharp phase transition point as the fundamental limit of pruning ratio.
AlphaPruning optimizes LLM pruning using HT-SR theory for better performance.
problem Improving pruning of large language models to reduce size without sacrificing performance.
method AlphaPruning uses HT-SR theory to allocate layerwise sparsity ratios more theoretically.
result AlphaPruning prunes LLaMA-7B to 80% sparsity with reasonable perplexity.
DSA efficiently allocates sparsity across layers for budgeted pruning.
problem Efficiently distributing resources (sparsity) across layers in pruning under resource constraints.
method DSA uses differentiable pruning to find continuous layer-wise pruning ratios via gradient-based optimization.
result DSA achieves superior performance and significantly reduces the time cost of pruning.
Deep Neural Network (DNN) is powerful but computationally expensive and memory intensive, thus impeding its practical usage on resource-constrained front-end devices. DNN pruning is an approach for deep model compression, which aims at eliminating some parameters with tolerable performance degradation. In this paper, w…
Alpha-trimming prunes trees in random forests to improve predictive performance.
problem Improving predictive performance of random forests by locally adaptive tree pruning.
method Alpha-trimming is a fast pruning algorithm that prunes trees in a random forest based on signal-to-noise ratio, controlled by a tuning parameter.
result Alpha-trimming often lowers mean squared prediction error compared to fully grown random forests.
Randomly initialized networks can perform as well as pruned networks.
problem Understanding and improving network pruning methods.
method Sanity checks on recent pruning methods, proposing random tickets.
result Randomly initialized networks can perform as well as pruned networks.
Compression techniques for deep neural networks are important for implementing them on small embedded devices. In particular, channel-pruning is a useful technique for realizing compact networks. However, many conventional methods require manual setting of compression ratios in each layer. It is difficult to analyze th…
Deep neural networks (DNNs) although achieving human-level performance in many domains, have very large model size that hinders their broader applications on edge computing devices. Extensive research work have been conducted on DNN model compression or pruning. However, most of the previous work took heuristic approac…
BWS selects best window subsets for efficient data pruning.
problem Challenges in selecting subsets of large datasets for neural network training.
method Best Window Selection (BWS) by choosing optimal window intervals from ordered sample scores.
result BWS outperforms other methods across various selection ratios and datasets.
Pruning is an efficient model compression technique to remove redundancy in the connectivity of deep neural networks (DNNs). Computations using sparse matrices obtained by pruning parameters, however, exhibit vastly different parallelism depending on the index representation scheme. As a result, fine-grained pruning ha…
Sparsity helps reduce the computational complexity of deep neural networks by skipping zeros. Taking advantage of sparsity is listed as a high priority in next generation DNN accelerators such as TPU. The structure of sparsity, i.e., the granularity of pruning, affects the efficiency of hardware accelerator design as w…
A new algorithm detects changes in data with constant cost per iteration.
problem Detecting changes in data with low computational cost.
method Adapting pruning and maximisation techniques from Gaussian data to exponential family models.
result The algorithm can detect changes in a wide range of models with a constant per-iteration cost.
Despite enjoying extensive applications in video analysis, three-dimensional convolutional neural networks (3D CNNs)are restricted by their massive computation and storage consumption. To solve this problem, we propose a threedimensional regularization-based neural network pruning method to assign different regularizat…
New compression theory justifies model pruning for neural networks.
problem Improving neural network performance with reduced model size.
method Information-theoretic rate-distortion theory applied to NN compression.
result Pruning improves model performance on CIFAR-10 and ImageNet datasets.
Deep Neural Networks (DNNs) are applied in a wide range of usecases. There is an increased demand for deploying DNNs on devices that do not have abundant resources such as memory and computation units. Recently, network compression through a variety of techniques such as pruning and quantization have been proposed to r…
Simple iterative method reduces deep network size significantly.
problem Overparameterization of deep neural networks in compute-limited systems.
method Hybrid approach combining single shot pruning and Lottery-Ticket methods.
result State-of-the-art compression achieved with improved test accuracy and compression ratio.
i-SpaSP prunes neural networks by identifying important groups of parameters, improving pruning efficiency.
problem Pruning neural networks to reduce computational cost and improve performance.
method i-SpaSP uses sparse signal recovery principles to iteratively identify and threshold important parameter groups.
result i-SpaSP achieves strong empirical results and theoretical convergence guarantees, improving pruning efficiency.
DJPQ optimizes neural network pruning and quantization for hardware efficiency.
problem Efficiently compress neural networks for hardware inference.
method Joint gradient-based optimization of pruning and quantization into a differentiable loss function.
result Significant reduction in Bit-Operations (BOPs) with minimal accuracy loss.
This paper finds a new way to compress CNN weights, improving on pruning and quantization.
problem Improving performance and storage efficiency of CNNs.
method Identifying and exploiting repeated patterns in CNN weight tensors, using Huffman coding and block sparse matrix formats.
result Achieved compaction ratios of 1.4x to 3.1x in addition to pruning and quantization.
This paper introduces blind adversarial pruning to balance accuracy, efficiency, and robustness in neural networks.
problem Balancing accuracy, efficiency, and robustness in neural networks with limited resources.
method Adversarial pruning with a cutoff-scale strategy to dynamically adjust the strength of adversarial examples.
result Blind adversarial pruning improves the overall AER of pruned models compared to adversarial pruning.
ResRep prunes CNNs without losing accuracy by separating remembering and forgetting.
problem Pruning CNNs to reduce FLOPs without sacrificing accuracy.
method Decoupling remembering and forgetting in CNNs, using SGD for remembering and a novel update rule for forgetting.
result Achieved lossless pruning with high compression ratio (76.15% accuracy on ImageNet with 45% FLOPs reduction).
Recently, pre-trained language representation flourishes as the mainstay of the natural language understanding community, e.g., BERT. These pre-trained language representations can create state-of-the-art results on a wide range of downstream tasks. Along with continuous significant performance improvement, the size an…
Two retraining techniques outperform fine-tuning in neural network pruning.
problem Improving accuracy and compression in neural network pruning.
method Weight rewinding and learning rate rewinding compared to fine-tuning.
result Rewinding techniques outperform fine-tuning in accuracy and compression.
This study argues for pruning trees in random forests to improve performance in low signal-to-noise scenarios.
problem Improving random forest performance in scenarios with low signal-to-noise ratio.
method Using regularization theory, the study re-examines the depth of trees in random forests and provides evidence that shallow trees are advantageous.
result Random forests with shallow trees are advantageous when the signal-to-noise ratio is low.
Auto-Compressing Subset Pruning reduces model size for faster inference.
problem High parameter counts and slow inference times in semantic segmentation models.
method Learning a channel selection mechanism based on temperature annealing schedule.
result Significant compression of segmentation models with acceptable inference performance.
Model compression techniques, such as pruning and quantization, are becoming increasingly important to reduce the memory footprints and the amount of computations. Despite model size reduction, achieving performance enhancement on devices is, however, still challenging mainly due to the irregular representations of spa…
A flexible nonparametric online changepoint detection algorithm for high-frequency data.
problem Detecting changes in real-time in high-frequency data streams with limited computational resources.
method NP-FOCuS, a sequential likelihood ratio test for a change in the empirical cumulative density function, using functional pruning.
result NP-FOCuS outperforms current nonparametric online changepoint techniques in various settings.
Kernel Quantization improves CNN compression without sacrificing performance.
problem Efficiently compressing CNN models without significant performance loss.
method Quantizes convolution kernels as the unit, learning a codebook for low-bit indexes.
result Significant compression ratio achieved with minimal accuracy loss.
SeReNe prunes neurons with low sensitivity to reduce network size.
problem Large neural networks consume too many resources on resource-constrained devices.
method Exploits neural sensitivity as a regularizer to prune neurons with low sensitivity.
result Pruning neurons with low sensitivity achieves competitive compression ratios.
Binary Stochastic Filtering (BSF), the algorithm for feature selection and neuron pruning is proposed in this work. The method defines filtering layer which penalizes amount of the information involved in the training process. This information could be the input data or output of the previous layer, which directly lead…
It is inevitable to train large deep learning models on a large-scale cluster equipped with accelerators system. Deep gradient compression would highly increase the bandwidth utilization and speed up the training process but hard to implement on ring structure. In this paper, we find that redundant gradient and gradien…
New study reveals how heavy-tailed SGD dynamics lead to compressible neural networks.
problem Understanding why large neural networks can be compressed effectively.
method Linking SGD dynamics to compressibility properties of neural networks.
result Large step-size/batch-size ratios and overparametrization lead to heavy-tailed SGD dynamics, making networks compressible.
Bayesian model averaging under predictor redundancy
problem Reporting Bayesian model averaging posterior without changing the Bayesian target
method Using hard or soft regions of support space
result Region reports often give shorter and clearer summaries while preserving the main posterior information
Model compression has been introduced to reduce the required hardware resources while maintaining the model accuracy. Lots of techniques for model compression, such as pruning, quantization, and low-rank approximation, have been suggested along with different inference implementation characteristics. Adopting model com…
Paper prunes deep MIR models to ultra-light versions.
problem Massive complexity of deep learning models in MIR.
method Lottery ticket hypothesis-based model pruning.
result Up to 90% of model parameters can be removed without loss of accuracy.
Recent pruning methods at initialization fall short of random pruning's accuracy.
problem Improving neural network accuracy through pruning at initialization.
method Various pruning methods (SNIP, GraSP, SynFlow, magnitude pruning) are evaluated; per-layer pruning decisions are proposed.
result Randomly shuffling or sampling initial weights preserves or improves accuracy, suggesting challenges with pruning heuristics.
ICE-Pruning accelerates deep neural network pruning by 9.61x.
problem Efficiently pruning deep neural networks while maintaining accuracy.
method Iterative pruning with automatic fine-tuning steps, freezing strategy, and custom learning rate scheduler.
result Significantly reduces pruning time by up to 9.61x.
Magnitude-based pruning is one of the simplest methods for pruning neural networks. Despite its simplicity, magnitude-based pruning and its variants demonstrated remarkable performances for pruning modern architectures. Based on the observation that magnitude-based pruning indeed minimizes the Frobenius distortion of a…
Pruning neural network parameters is often viewed as a means to compress models, but pruning has also been motivated by the desire to prevent overfitting. This motivation is particularly relevant given the perhaps surprising observation that a wide variety of pruning approaches increase test accuracy despite sometimes …
Large-scale deep neural networks (DNNs) are both compute and memory intensive. As the size of DNNs continues to grow, it is critical to improve the energy efficiency and performance while maintaining accuracy. For DNNs, the model size is an important factor affecting performance, scalability and energy efficiency. Weig…
Dynamic pruning during training reduces deep network complexity without significant accuracy loss.
problem High memory and computational requirements of deep networks during training and inference.
method Dynamic pruning of convolutional filters during training, using L1 normalization for optimization.
result L1 normalization-based pruning yields up to 50% reduction in filters with minimal accuracy loss.
The study reveals flaws in pruning criteria and proposes a new assumption for better filter selection.
problem Flaws in existing pruning criteria for CNNs.
method Empirical experiments and Convolutional Weight Distribution Assumption.
result The Convolutional Weight Distribution Assumption improves filter selection in pruning.
Deep neural networks have dramatically achieved great success on a variety of challenging tasks. However, most successful DNNs have an extremely complex structure, leading to extensive research on model compression.As a significant area of progress in model compression, traditional gradual pruning approaches involve an…
Gibbs pruning optimizes neural networks by combining physics and regularization.
problem Large neural networks are impractical for many applications.
method Combines statistical physics and stochastic regularization to train and prune networks simultaneously.
result Gibbs pruning achieves state-of-the-art performance on ResNet-56.
We examine how recently documented, fundamental phenomena in deep learning models subject to pruning are affected by changes in the pruning procedure. Specifically, we analyze differences in the connectivity structure and learning dynamics of pruned models found through a set of common iterative pruning techniques, to …
New statistical mechanics analysis shows edge pruning outperforms node pruning in neural networks.
problem Theoretical understanding of neural network pruning effectiveness is lacking.
method Statistical mechanics analysis of a teacher-student framework.
result DPP node pruning method is superior to other methods, but edge pruning is better overall.
New algorithm PRACTISE accelerates networks with tiny sets, reducing latency by 22%.
problem Accelerating networks with tiny training sets to improve performance and privacy.
method Proposes a new algorithm PRACTISE based on finetune convexity hypothesis and feature mimicking.
result PRACTISE outperforms previous methods by reducing latency by 22% on ImageNet-1k.
Temporal VAE improves VaR estimation for financial portfolios.
problem Estimating VaR for large asset portfolios in finance.
method Temporal VAE with annealing regularization to avoid posterior collapse.
result Temporal VAE outperforms classical VaR estimation methods on real data.