English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In this paper, we start by training End-to-End Automatic Speech Recognition(ASR) models using Federated Learning (FL) and examining the fundamentalconsiderations that can be pivotal in minimizing the performance gap in termsof word error rate between models trained using FL versus their centralizedcounterpart. Specifically, we study the effect of (i) adaptive optimizers, (ii)loss characteristics via altering Connectionist Temporal Classification (CTC)weight, (iii) model initialization through seed start, (iv) carrying overmodeling setup from experiences in centralized training to FL, e.g., pre-layeror post-layer normalization, and (v) FL-specific hyperparameters, such asnumber of local epochs, client sampling size, and learning rate scheduler,specifically for ASR under heterogeneous data distribution. We shed light onhow some optimizers work better than others via inducing smoothness. We alsosummarize the applicability of algorithms, trends, and propose best practicesfrom prior works in FL (in general) toward End-to-End ASR models.

Importance of Smoothness Induced by Optimizers in FL4ASR: Towards Understanding Federated Learning for End-to-End ASR