Training Keyword Spotting Models on Non-IID Data with Federated Learning

• Main Authors: Hard, Andrew, Partridge, Kurt, Nguyen, Cameron, Subrahmanya, Niranjan, Shah, Aishanee, Zhu, Pai, Moreno, Ignacio Lopez, Mathews, Rajiv
• Format: Journal Article
• Language: English
• Published: 20-05-2020
• Subjects: Computer Science - Computation and Language; Computer Science - Learning; Computer Science - Sound
• DOI: 10.48550/arxiv.2005.10406

We demonstrate that a production-quality keyword-spotting model can be
trained on-device using federated learning and achieve comparable false accept
and false reject rates to a centrally-trained model. To overcome the
algorithmic constraints associated with fitting on-device data (which are
inherently non-independent and identically distributed), we conduct thorough
empirical studies of optimization algorithms and hyperparameter configurations
using large-scale federated simulations. To overcome resource constraints, we
replace memory intensive MTR data augmentation with SpecAugment, which reduces
the false reject rate by 56%. Finally, to label examples (given the zero
visibility into on-device data), we explore teacher-student training.