No abstract

No abstract

Traditional Federated Learning (FL) adopts a clientserver architecture where FL clients (e.g. IoT edge devices) train a common global model with the help of a centralised orchestrator (cloud server). However, current approaches are moving away from centralised orchestration towards a decentralised one in order to fully adapt FL for a cross-silo configuration with multiple organisations acting as clients. State-of-the-art decentralised FL mechanisms make at least one of the following assumptions: (a) clients are trusted organisations and cannot inject low quality model updates for aggregation and (b) client local models can be shared with other clients or a third party for verification of low quality updates. This paper proposes a Blockchain based decentralised framework for scenarios where participatory organisations are believed to be fully capable of injecting low quality model updates as they are not willing to expose their local models to any other entity for verification purpose. The proposed decentralised FL framework adopts a novel hierarchical network of aggregators with the ability to punish/reward organisations in proportion to their local model quality updates. The framework is flexible and unlike state-ofthe-art solutions, prevents a single entity from possessing the aggregated model in any FL round of training. The proposed framework is tested with respect to off-chain and on-chain performance in two Industry 4.0 use-cases: predictive maintenance and product visual inspection. A comparative evaluation against the state-of-the-art reveals the proposed framework's utility in terms of minimizing model convergence time and latency while maximizing accuracy and throughput.

No abstract

No abstract

No abstract