在设备端同时调整模型权重和结构，实现高效个性化关键词识别。
OnDA: On-device Channel Pruning for Efficient Personalized Keyword Spotting
- 结合在线结构化通道剪枝与权重自适应，动态优化模型结构。
- 在保持0.5次/小时误报率下，模型大小压缩至原模型的1/9.63。
- 适合资源受限场景下的实时个性化语音识别应用。
始终开启的关键词识别（KWS）需要在严苛的延迟和能耗约束下，对用户及环境变化进行设备端自适应。本文首次提出将权重自适应（即设备端训练）与架构自适应相结合，通过在线结构化通道剪枝实现个性化设备端KWS。基于先进的自学习个性化KWS流程，我们对比了无数据依赖与有数据感知的剪枝策略，应用于现场伪标签用户数据。在HeySnips和HeySnapdragon数据集上，相对于未剪枝基线，在保持相同任务性能（0.5次/小时误报率）条件下，模型尺寸最大压缩9.63倍。在Jetson Orin Nano嵌入式GPU上部署该自适应流程后，相较于仅权重调整方案，在线训练与推理阶段分别实现最高1.52倍/1.57倍和1.64倍/1.77倍的延迟与能耗提升。
原文摘要 · Abstract (English)
Always-on keyword spotting (KWS) demands on-device adaptation to cope with user- and environment-specific distribution shifts under tight latency and energy budgets. This paper proposes, for the first time, coupling weight adaptation (i.e., on-device training) with architectural adaptation, in the form of online structured channel pruning, for personalized on-device KWS. Starting from a state-of-the-art self-learning personalized KWS pipeline, we compare data-agnostic and data-aware pruning criteria applied on in-field pseudo-labelled user data. On the HeySnips and HeySnapdragon datasets, we achieve up to 9.63x model-size compression with respect to unpruned baselines at iso-task performance, measured as the accuracy at 0.5 false alarms per hour. When deploying our adaptation pipeline on a Jetson Orin Nano embedded GPU, we achieve up to 1.52x/1.57x and 1.64x/1.77x latency and energy-consumption improvements during online training/inference compared to weights-only adaptation.
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。