OnDA: On-device Channel Pruning for Efficient Personalized Keyword Spotting
Not on the map. Why?
Abstract (arXiv)
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.
Key facts. No citing papers in the map yet (data through 13 Oct 2026). Builds on 16 of its 21 references that are in the map. First posted on arXiv 25 Feb 2026.
Builds on
16 of 21 references.
References · 16
- 2017Prototypical Networks for Few-shot Learningcited by 3,011
- 2016Pruning Filters for Efficient ConvNetscited by 1,138
- 2021Sparsity in Deep Learning: Pruning and growth for efficient inference and training in neural networkscited by 525
- 2017Hello Edge: Keyword Spotting on Microcontrollerscited by 108
- 2017Deep Residual Learning for Small-Footprint Keyword Spottingcited by 68
- 2021Hessian-Aware Pruning and Optimal Neural Implantcited by 45
- 2021Deep Spoken Keyword Spotting: An Overviewcited by 43
- 2018Efficient keyword spotting using dilated convolutions and gatingcited by 21
- 2019Query-by-example on-device keyword spottingcited by 19
- 2020Few-Shot Keyword Spotting With Prototypical Networkscited by 18
- 2023Lightweight Neural Architecture Search for Temporal Convolutional Networks at the Edgecited by 15
- 2024On-Device Domain Learning for Keyword Spotting on Low-Power Extreme Edge Embedded Systemscited by 13
4 more
- 2023Few-Shot Open-Set Learning for On-Device Customization of KeyWord Spotting Systemscited by 9
- 2022Personalized Keyword Spotting through Multi-task Learningcited by 7
- 2024Self-Learning for Personalized Keyword Spotting on Ultra-Low-Power Audio Sensorscited by 7
- 2022Keyword Spotting System and Evaluation of Pruning and Quantization Methods on Low-power Edge Microcontrollerscited by 4
Built on by
Not cited yet.