Paper Title: RCT: Rewarding Clients Tool in Decentralized Carbon Emission Classification
Conference Name: KICS Summer 2024
Abstract: Rewarding Clients Tool, RCT, integrates methods to predict carbon emissions and encourage client participation, by using a two-layer model, Federated Learning Layer and the Rewarding Layer. Local computing clients in the Federated Learning Layer, training models on local data securely stored in IPFS. The aggregating client combines models using Federated Averaging, maintaining data privacy. A blockchain-based smart contract governs the process, ensuring transparency and overall flow. The Rewarding Layer distributes incentive to clients based on model accuracy, motivating participation. RCT's architecture yields a precise global CO2 emission model, aiding in emission reduction efforts policymakers while upholding data security.