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/

Zendy Plus

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

Zendy Tools

Zendy Open

How can we debug a logistical regression model in a federated learningsetting when seeing the model behave unexpectedly (e.g., the model rejects allhigh-income customers' loan applications)? The SQL-based training datadebugging framework has proved effective to fix this kind of issue in anon-federated learning setting. Given an unexpected query result over modelpredictions, this framework automatically removes the label errors fromtraining data such that the unexpected behavior disappears in the retrainedmodel. In this paper, we enable this powerful framework for federated learning.The key challenge is how to develop a security protocol for federated debuggingwhich is proved to be secure, efficient, and accurate. Achieving this goalrequires us to investigate how to seamlessly integrate the techniques frommultiple fields (Databases, Machine Learning, and Cybersecurity). We firstpropose FedRain, which extends Rain, the state-of-the-art SQL-based trainingdata debugging framework, to our federated learning setting. We address severaltechnical challenges to make FedRain work and analyze its security guaranteeand time complexity. The analysis results show that FedRain falls short interms of both efficiency and security. To overcome these limitations, weredesign our security protocol and propose Frog, a novel SQL-based trainingdata debugging framework tailored for federated learning. Our theoreticalanalysis shows that Frog is more secure, more accurate, and more efficient thanFedRain. We conduct extensive experiments using several real-world datasets anda case study. The experimental results are consistent with our theoreticalanalysis and validate the effectiveness of Frog in practice.

Enabling SQL-based Training Data Debugging for Federated Learning