Pruning as Regularization: Sensitivity-Aware One-Shot Pruning in ASR

• URL: http://arxiv.org/abs/2511.08092v1
• Date: Wed, 12 Nov 2025 01:39:09 GMT
• Title: Pruning as Regularization: Sensitivity-Aware One-Shot Pruning in ASR
• Abstract summary: We demonstrate that one-shot magnitude pruning serves as a powerful implicit regularizer for ASR.<n>We show that decoder FFNs are pruning-fragile, whereas decoder self-attention and the last encoder layers contain redundancy that, when removed, improves generalization.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We challenge the conventional view of neural network pruning as solely a compression technique, demonstrating that one-shot magnitude pruning serves as a powerful implicit regularizer for ASR. Using Whisper-small, we combine gradient- and Fisher-based sensitivity diagnostics with targeted, component-wise pruning. This reveals architectural asymmetries: decoder FFNs are pruning-fragile, whereas decoder self-attention and the last encoder layers contain redundancy that, when removed, improves generalization. Without fine-tuning, pruning 50% of decoder self-attention reduces WER by 2.38% absolute (20.44% relative) on LibriSpeech test-other; pruning the last four encoder layers at 50% instead yields a 1.72% absolute (14.8% relative) improvement. Gains persisted on Common Voice and TED-LIUM datasets. Beyond regularization benefits, our sensitivity-aware approach enables more aggressive one-shot compression. At 40% sparsity, where established global pruning approaches catastrophically fail, our method preserves near-baseline accuracy. This positions pruning as a first-class architectural design tool: knowing where to prune is as important as how much to prune.

Related papers

• ARMOR: High-Performance Semi-Structured Pruning via Adaptive Matrix Factorization [99.96330641363396]
  ARMOR: (Adaptive Representation with Matrix-factORization) is a novel one-shot post-training pruning algorithm.<n>Instead of directly pruning weights, ARMOR factorizes each weight matrix into a 2:4 sparse core wrapped by two low-overhead, block diagonal matrices.<n>We demonstrate ARMOR consistently and significantly outperforms state-of-the-art 2:4 pruning methods across a wide range of downstream tasks and perplexity evaluations.
  arXiv  Detail & Related papers  (2025-10-07T02:39:20Z)
• Patch-Level Contrasting without Patch Correspondence for Accurate and Dense Contrastive Representation Learning [79.43940012723539]
  ADCLR is a self-supervised learning framework for learning accurate and dense vision representation. Our approach achieves new state-of-the-art performance for contrastive methods.
  arXiv  Detail & Related papers  (2023-06-23T07:38:09Z)
• DiffRate : Differentiable Compression Rate for Efficient Vision Transformers [98.33906104846386]
  Token compression aims to speed up large-scale vision transformers (e.g. ViTs) by pruning (dropping) or merging tokens. DiffRate is a novel token compression method that has several appealing properties prior arts do not have.
  arXiv  Detail & Related papers  (2023-05-29T10:15:19Z)
• Vision Transformer Compression with Structured Pruning and Low Rank Approximation [1.9685957565449135]
  Transformer architecture has gained popularity due to its ability to scale with large dataset. We focus on vision transformer proposed for image recognition task. We explore the application of different compression techniques such as low rank approximation and pruning for this purpose.
  arXiv  Detail & Related papers  (2022-03-25T04:18:07Z)
• Towards Compact CNNs via Collaborative Compression [166.86915086497433]
  We propose a Collaborative Compression scheme, which joints channel pruning and tensor decomposition to compress CNN models. We achieve 52.9% FLOPs reduction by removing 48.4% parameters on ResNet-50 with only a Top-1 accuracy drop of 0.56% on ImageNet 2012.
  arXiv  Detail & Related papers  (2021-05-24T12:07:38Z)
• Toward Compact Deep Neural Networks via Energy-Aware Pruning [2.578242050187029]
  We propose a novel energy-aware pruning method that quantifies the importance of each filter in the network using nuclear-norm (NN) We achieve competitive results with 40.4/49.8% of FLOPs and 45.9/52.9% of parameter reduction with 94.13/94.61% in the Top-1 accuracy with ResNet-56/110 on CIFAR-10.
  arXiv  Detail & Related papers  (2021-03-19T15:33:16Z)
• Hessian-Aware Pruning and Optimal Neural Implant [74.3282611517773]
  Pruning is an effective method to reduce the memory footprint and FLOPs associated with neural network models. We introduce a new Hessian Aware Pruning method coupled with a Neural Implant approach that uses second-order sensitivity as a metric for structured pruning.
  arXiv  Detail & Related papers  (2021-01-22T04:08:03Z)
• Non-Parametric Adaptive Network Pruning [125.4414216272874]
  We introduce non-parametric modeling to simplify the algorithm design. Inspired by the face recognition community, we use a message passing algorithm to obtain an adaptive number of exemplars. EPruner breaks the dependency on the training data in determining the "important" filters.
  arXiv  Detail & Related papers  (2021-01-20T06:18:38Z)
• KCP: Kernel Cluster Pruning for Dense Labeling Neural Networks [5.274653527674298]
  We propose kernel cluster pruning (KCP) to prune dense labeling networks. Our method can reduce more than 70% of FLOPs with less than 1% of accuracy drop. For ResNet-50 on ILSVRC-2012, our KCP can reduce more than 50% of FLOPs reduction with 0.13% Top-1 accuracy gain.
  arXiv  Detail & Related papers  (2021-01-17T14:59:00Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.