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

Billions of distributed, heterogeneous and resource constrained IoT devicesdeploy on-device machine learning (ML) for private, fast and offline inferenceon personal data. On-device ML is highly context dependent, and sensitive touser, usage, hardware and environment attributes. This sensitivity and thepropensity towards bias in ML makes it important to study bias in on-devicesettings. Our study is one of the first investigations of bias in this emergingdomain, and lays important foundations for building fairer on-device ML. Weapply a software engineering lens, investigating the propagation of biasthrough design choices in on-device ML workflows. We first identify reliabilitybias as a source of unfairness and propose a measure to quantify it. We thenconduct empirical experiments for a keyword spotting task to show how complexand interacting technical design choices amplify and propagate reliabilitybias. Our results validate that design choices made during model training, likethe sample rate and input feature type, and choices made to optimize models,like light-weight architectures, the pruning learning rate and pruningsparsity, can result in disparate predictive performance across male and femalegroups. Based on our findings we suggest low effort strategies for engineers tomitigate bias in on-device ML.

Tiny, always-on and fragile: Bias propagation through design choices in on-device machine learning workflows