Federated Learning for Keyword Spotting

Luque, David; Coucke, Alice; Lavril, Thibaut; Gisselbrecht, Thibault; Dureau, Joseph

doi:10.48550/arxiv.1810.05512

Cited by 7 publications

(14 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…voice commands in [15], and proprietary data by Huawei in [4]; and (3) realistic federated datasets that are derived from publicly available data, but which are not straightforward to reproduce, e.g., FaceScrub in [19], Shakespeare in [16] and Reddit in [10,18,3].…”

Section: Reference Implementations Metricsmentioning

confidence: 99%

LEAF: A Benchmark for Federated Settings

Caldas¹,

Duddu²,

Wu³

et al. 2018

Preprint

269

480

View full text Add to dashboard Cite

Modern federated networks, such as those comprised of wearable devices, mobile phones, or autonomous vehicles, generate massive amounts of data each day. This wealth of data can help to learn models that can improve the user experience on each device. However, learning in federated settings presents new challenges at all stages of the machine learning pipeline. As the machine learning community begins to tackle these challenges, we are at a critical time to ensure that developments made in this area are grounded in real-world assumptions. To this end, we propose LEAF, a modular benchmarking framework for learning in federated settings. LEAF includes a suite of open-source federated datasets, a rigorous evaluation framework, and a set of reference implementations, all geared towards capturing the obstacles and intricacies of practical federated environments.

show abstract

Section: Reference Implementations Metricsmentioning

confidence: 99%

LEAF: A Benchmark for Federated Settings

Caldas¹,

Duddu²,

Wu³

et al. 2018

Preprint

269

480

View full text Add to dashboard Cite

show abstract

“…FedAvg is shown to generalize well while improving performance in terms of speed-ups. Lately, "Generalized FedAvg" was presented in (Reddi et al, 2020;Leroy et al, 2019), with similarities to the proposed hierarchical optimization method.…”

Section: Federated Transfer Learning (Ftl) Platformmentioning

confidence: 99%

“…has not been applied before -albeit, some work exists for KWS (Leroy et al, 2019). The e2e models have gained in popularity in ASR tasks because acoustic, language, and pronunciation models of a conventional ASR system can be combined into a single neural network (Chiu et al, 2018), including "Recurrent Neural Network Transducer" (RNN-T) (Graves, 2012), "Listen, Attend and Spell" (LAS) (Chan et al, 2015) and others.…”

Section: Introduction -Prior Workmentioning

confidence: 99%

Dynamic Gradient Aggregation for Federated Domain Adaptation

Dimitriadis¹,

Kumatani²,

Gmyr³

et al. 2021

Preprint

View full text Add to dashboard Cite

In this paper, a new learning algorithm for Federated Learning (FL) is introduced. The proposed scheme is based on a weighted gradient aggregation using two-step optimization to offer a flexible training pipeline. Herein, two different flavors of the aggregation method are presented, leading to an order of magnitude improvement in convergence speed compared to other distributed or FL training algorithms like BMUF and FedAvg. Further, the aggregation algorithm acts as a regularizer of the gradient quality. We investigate the effect of our FL algorithm in supervised and unsupervised Speech Recognition (SR) scenarios. The experimental validation is performed based on three tasks: first, the LibriSpeech task showing a speed-up of 7× and 6% word error rate reduction (WERR) compared to the baseline results. The second task is based on session adaptation providing 20% WERR over a powerful LAS model. Finally, our unsupervised pipeline is applied to the conversational SR task. The proposed FL system outperforms the baseline systems in both convergence speed and overall model performance.A massively distributed FL training for Automatic Speech Recognition (ASR) models, like the one herein presented,

show abstract

“…FedAvg is shown to generalize well while significantly improving performance in terms of speedups. Lately, "Generalized FedAvg" was presented in [15,11], a method with similarities to our proposed hierarchical optimization method.…”

Section: Hierarchical Optimizationmentioning

confidence: 99%

“…To the best of our knowledge, a massively distributed and heterogeneous approach, like the one herein presented for Automatic Speech Recognition (ASR), has not been applied before -albeit, some work exists for KWS [11]. An end-to-end (e2e) architecture is implemented on the Federated Learning platform for this particular SR task.…”

Section: Introductionmentioning

confidence: 99%

Federated Transfer Learning with Dynamic Gradient Aggregation

Dimitriadis¹,

Kumatani²,

Gmyr³

et al. 2020

Preprint

View full text Add to dashboard Cite

In this paper, a Federated Learning (FL) simulation platform is introduced. The target scenario is Acoustic Model training based on this platform. To our knowledge, this is the first attempt to apply FL techniques to Speech Recognition tasks due to the inherent complexity. The proposed FL platform can support different tasks based on the adopted modular design. As part of the platform, a novel hierarchical optimization scheme and two gradient aggregation methods are proposed, leading to almost an order of magnitude improvement in training convergence speed compared to other distributed or FL training algorithms like BMUF and FedAvg. The hierarchical optimization offers additional flexibility in the training pipeline besides the enhanced convergence speed. On top of the hierarchical optimization, a dynamic gradient aggregation algorithm is proposed, based on a data-driven weight inference. This aggregation algorithm acts as a regularizer of the gradient quality. Finally, an unsupervised training pipeline tailored to FL is presented as a separate training scenario. The experimental validation of the proposed system is based on two tasks: first, the LibriSpeech task showing a speed-up of 7× and 6% word error rate reduction (WERR) compared to the baseline results. The second task is based on session adaptation providing an improvement of 20% WERR over a competitive production-ready LAS model. The proposed Federated Learning system is shown to outperform the golden standard of distributed training in both convergence speed and overall model performance.

show abstract

Federated Learning for Keyword Spotting

Cited by 7 publications

References 0 publications

LEAF: A Benchmark for Federated Settings

LEAF: A Benchmark for Federated Settings

Dynamic Gradient Aggregation for Federated Domain Adaptation

Federated Transfer Learning with Dynamic Gradient Aggregation

Contact Info

Product

Resources

About