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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

The Federated Learning (FL) paradigm faces several challenges that limit itsapplication in real-world environments. These challenges include the localmodels' architecture heterogeneity and the unavailability of distributedInternet of Things (IoT) nodes due to connectivity problems. These factorsposit the question of "how can the available models fill the training gap ofthe unavailable models?". This question is referred to as the "ObliqueFederated Learning" problem. This problem is encountered in the studiedenvironment that includes distributed IoT nodes responsible for predicting CO2concentrations. This paper proposes the Correlation-based FL (CorrFL) approachinfluenced by the representational learning field to address this problem.CorrFL projects the various model weights to a common latent space to addressthe model heterogeneity. Its loss function minimizes the reconstruction losswhen models are absent and maximizes the correlation between the generatedmodels. The latter factor is critical because of the intersection of thefeature spaces of the IoT devices. CorrFL is evaluated on a realistic use case,involving the unavailability of one IoT device and heightened activity levelsthat reflect occupancy. The generated CorrFL models for the unavailable IoTdevice from the available ones trained on the new environment are comparedagainst models trained on different use cases, referred to as the benchmarkmodel. The evaluation criteria combine the mean absolute error (MAE) ofpredictions and the impact of the amount of exchanged data on the predictionperformance improvement. Through a comprehensive experimental procedure, theCorrFL model outperformed the benchmark model in every criterion.

CorrFL: Correlation-Based Neural Network Architecture for Unavailability Concerns in a Heterogeneous IoT Environment