pruning不仅能压缩模型，还能提升语音识别效果。
Pruning as Regularization: Sensitivity-Aware One-Shot Pruning in ASR
- 根据敏感度分析，有针对性地剪枝不同组件
- 剪掉解码器自注意力层50%后，测试集错误率降2.38%
- 适合想在不微调情况下提升模型泛化能力的研究者
我们挑战了神经网络剪枝仅用于压缩的传统观念，证明一次性幅度剪枝在语音识别中可作为强大的隐式正则化手段。以Whisper-small为例，结合梯度与Fisher信息敏感度诊断，对不同组件进行针对性剪枝。结果显示：解码器前馈网络对剪枝敏感，而解码器自注意力和最后一个编码器层存在冗余，剪除后能提升泛化性能。未微调情况下，剪掉解码器自注意力50%使LibriSpeech test-other的词错误率（WER）绝对降低2.38%（相对降低20.44%）；剪掉最后四个编码器层50%时，绝对降低1.72%（相对降低14.8%）。该效果在Common Voice和TED-LIUM数据集上也持续存在。此外，该敏感度感知方法可在40%稀疏度下保持接近基线的准确率，而传统全局剪枝在此处已崩溃。这表明剪枝应被视为首要的架构设计工具：知道在哪剪，比剪多少更重要。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。