Effective Communication Compression Framework Using Federated Learning Privacy Model for Smart Grid
Abstract
The rapidly increasing number of electricity usage in smart grid is generating huge quantities of data in end devices. The generated electricity data has benign data and false data and these are collectively stored in local servers. To maintain data privacy on local data the Federated learning framework is approached. The divided local datasets are processed by local model to generate local parameters. All these local model’s parameters are communicated to the global model with digital sovereignty, compression for aggregation. Global model performs the common modelling and generates the global parameter; this is sent to the local model to update their parameter for the next iteration of modelling. The challenge in increased electricity data creates the complexity in communicating the local model parameter to the global model. The global model accuracy is maintained same when local model parameters are sparsified and quantized again in global model parameters are dequantized. This work proposes the effective communication compression framework with FedAvg Framework (ECC-FedAvg) with compression techniques, which increases the communication effectiveness and increases the communication efficiency in smart grid.
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.