Federated Learning and Decentralized Compute

Federated Learning (FL) enables the training of a global AI model by utilizing decentralized edge devices that perform local training on their datasets. In traditional FL architectures, a central server aggregates updates from these devices, posing risks such as a single point of failure and potential exposure of private data. To address these issues, Coven AI integrates Blockchain technology into the FL framework to decentralize the aggregation process and enhance security.

By leveraging Blockchain as a secure communication layer, updates from edge devices are encrypted and stored in a decentralized manner. This eliminates the dependence on a central server and ensures that no single entity has control over the aggregation process, mitigating risks of malicious activity and data reconstruction.